[WIP] Trying to fix issue

.github/actions/get-gltf-assets/action.yml

@@ -1,23 +0,0 @@
-name: 'Get and cache glTF Assets'
-description: 'Downloads and caches glTF assets by calling the get-gltf-sample-assets.sh script.'
-
-runs:
-  using: "composite"
-  steps:
-    - uses: ./.github/actions/dep-versions
-    - name: Hash models file
-      id: hash-models
-      shell: bash
-      run: echo "hash=$(cat test/renderdiff/tests/gltf_models.txt | md5sum | sed 's/ -//g')" >> $GITHUB_OUTPUT
-    - name: Cache glTF assets
-      id: cache-gltf
-      uses: actions/cache@v4
-      with:
-        path: gltf
-        key: gltf-assets-${{ env.GITHUB_GLTF_SAMPLE_ASSETS_COMMIT }}-${{ steps.hash-models.outputs.hash }}
-    - name: Download assets via script if cache not found
-      if: steps.cache-gltf.outputs.cache-hit != 'true'
-      shell: bash
-      run: |
-        echo "Cache miss for commit ${{ env.GITHUB_GLTF_SAMPLE_ASSETS_COMMIT }}. Running download script..."
-        xargs bash build/common/get-gltf-sample-assets.sh < test/renderdiff/tests/gltf_models.txt

.github/actions/get-mesa/action.yml

@@ -1,6 +1,5 @@
-name: 'Get and cache Mesa'
-description: 'Get and cache Mesa'
-
+name: 'Get Mesa'
+description: 'Caches and installs Mesa'
 runs:
   using: "composite"
   steps:

.github/actions/get-vulkan-sdk/action.yml

@@ -8,8 +8,8 @@ runs:
       uses: actions/cache@v3
       id: cache-vulkan-sdk
       with:
-        path: ${{ runner.homedir }}/VulkanSDK
-        key: vulkansdk-${{ env.GITHUB_VULKANSDK_VERSION }}-2-${{ runner.os }}
+        path: vulkansdk
+        key: vulkansdk-${{ env.GITHUB_VULKANSDK_VERSION }}-${{ runner.os }}
     - name: Download Vulkan SDK
       if: steps.cache-vulkan-sdk.outputs.cache-hit != 'true'
       run: |
@@ -21,8 +21,4 @@ runs:
       run: |
         source ${{ github.workspace }}/build/common/get-vulkan-sdk.sh
         unpack_vulkan_installer
-        pushd .
-        cd ~/VulkanSDK/${GITHUB_VULKANSDK_VERSION}
-        sudo ./install_vulkan.py
-        popd
       shell: bash

.github/actions/mac-prereq/action.yml

@@ -19,7 +19,5 @@ runs:
     - name: Install Mac Prerequisites
       shell: bash
       run: |
-        # Install brew prereqs
-        brew install coreutils
         # Install ninja
         source ./build/common/get-ninja.sh

.github/workflows/presubmit.yml

@@ -9,36 +9,6 @@ on:
       - main
 
 jobs:
-  build-desktop-mac:
-    name: build-mac
-    runs-on: macos-14-xlarge
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/mac-prereq
-      - name: Run build script
-        run: |
-          cd build/mac && printf "y" | ./build.sh presubmit
-      - name: Test material parser
-        run: |
-          out/cmake-release/filament/test/test_material_parser
-
-  build-desktop-linux:
-    name: build-linux
-    runs-on: ubuntu-22.04-16core
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/linux-prereq
-      - name: Run build script
-        run: |
-          cd build/linux && printf "y" | ./build.sh presubmit
-      - name: Test material parser
-        run: |
-          out/cmake-release/filament/test/test_material_parser
-
   build-windows:
     name: build-windows
     runs-on: windows-2022-32core
@@ -48,157 +18,194 @@ jobs:
           fetch-depth: 0
       - name: Run build script
         run: |
-          build\windows\build-github.bat presubmit
+          build\windows\build-github.bat continuous
         shell: cmd
 
-  build-android:
-    name: build-android
-    runs-on: 'ubuntu-24.04-16core'
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/linux-prereq
-      - uses: actions/setup-java@v3
-        with:
-          distribution: 'temurin'
-          java-version: '17'
-      - name: Run build script
-        # Only build 1 64 bit target during presubmit to cut down build times during presubmit
-        # Continuous builds will build everything
-        run: |
-          cd build/android && printf "y" | ./build.sh presubmit arm64-v8a
+  # build-desktop-mac:
+  #   name: build-mac
+  #   runs-on: macos-14-xlarge
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/mac-prereq
+  #     - name: Run build script
+  #       run: |
+  #         cd build/mac && printf "y" | ./build.sh presubmit
+  #     - name: Test material parser
+  #       run: |
+  #         out/cmake-release/filament/test/test_material_parser
 
-  build-ios:
-    name: build-iOS
-    runs-on: macos-14-xlarge
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/mac-prereq
-      - name: Run build script
-        run: |
-          cd build/ios && printf "y" | ./build.sh presubmit
-      - name: Build iOS samples
-        run: |
-          cd build/ios && ./build-samples.sh presubmit
+  # build-desktop-linux:
+  #   name: build-linux
+  #   runs-on: ubuntu-22.04-16core
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/linux-prereq
+  #     - name: Run build script
+  #       run: |
+  #         cd build/linux && printf "y" | ./build.sh presubmit
+  #     - name: Test material parser
+  #       run: |
+  #         out/cmake-release/filament/test/test_material_parser
 
-  build-web:
-    name: build-web
-    runs-on: 'ubuntu-24.04-16core'
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/linux-prereq
-      - uses: ./.github/actions/web-prereq
-      - name: Run build script
-        run: |
-          cd build/web && printf "y" | ./build.sh presubmit
+  # build-windows:
+  #   name: build-windows
+  #   runs-on: windows-2022-32core
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - name: Run build script
+  #       run: |
+  #         build\windows\build-github.bat presubmit
+  #       shell: cmd
 
-  validate-docs:
-    name: validate-docs
-    runs-on: 'ubuntu-24.04-4core'
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - id: get_commit_msg
-        uses: ./.github/actions/get-commit-msg
-      - name: Check for manual edits to /docs
-        run: |
-          COMMIT_ID=$(echo "${{ steps.get_commit_msg.outputs.msg }}" | head -n 1 | sed "s/commit //g")
-          bash docs_src/build/presubmit_check.sh ${COMMIT_ID}
+  # build-android:
+  #   name: build-android
+  #   runs-on: 'ubuntu-24.04-16core'
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/linux-prereq
+  #     - uses: actions/setup-java@v3
+  #       with:
+  #         distribution: 'temurin'
+  #         java-version: '17'
+  #     - name: Run build script
+  #       # Only build 1 64 bit target during presubmit to cut down build times during presubmit
+  #       # Continuous builds will build everything
+  #       run: |
+  #         cd build/android && printf "y" | ./build.sh presubmit arm64-v8a
 
-  test-renderdiff:
-    name: test-renderdiff
-    runs-on: macos-14-xlarge
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/mac-prereq
-      - uses: ./.github/actions/get-gltf-assets
-      - uses: ./.github/actions/get-mesa
-      - uses: ./.github/actions/get-vulkan-sdk      
-      - id: get_commit_msg
-        uses: ./.github/actions/get-commit-msg
-      - name: Prerequisites
-        run: |
-          pip install tifffile numpy
-          # Must have at least clang-16 for a webgpu/dawn build.
-          sudo xcode-select -s /Applications/Xcode_16.2.app/Contents/Developer
-        shell: bash
-      - name: Render and compare
-        id: render_compare
-        run: |
-          ls ./gltf/Models
-          TEST_DIR=test/renderdiff
-          source ${TEST_DIR}/src/preamble.sh
-          start_
-          GOLDEN_BRANCH=$(echo "${{ steps.get_commit_msg.outputs.msg }}" | python3 ${TEST_DIR}/src/commit_msg.py)
-          bash ${TEST_DIR}/generate.sh && \
-              python3 ${TEST_DIR}/src/golden_manager.py \
-                  --branch=${GOLDEN_BRANCH} \
-                  --output=${GOLDEN_OUTPUT_DIR}
+  # build-ios:
+  #   name: build-iOS
+  #   runs-on: macos-14-xlarge
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/mac-prereq
+  #     - name: Run build script
+  #       run: |
+  #         cd build/ios && printf "y" | ./build.sh presubmit
+  #     - name: Build iOS samples
+  #       run: |
+  #         cd build/ios && ./build-samples.sh presubmit
 
-          # Note that we need to upload the output even if comparison fails, so we undo `set -ex`
-          end_
+  # build-web:
+  #   name: build-web
+  #   runs-on: 'ubuntu-24.04-16core'
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/linux-prereq
+  #     - uses: ./.github/actions/web-prereq
+  #     - name: Run build script
+  #       run: |
+  #         cd build/web && printf "y" | ./build.sh presubmit
 
-          python3 ${TEST_DIR}/src/compare.py \
-              --src=${GOLDEN_OUTPUT_DIR} \
-              --dest=${RENDER_OUTPUT_DIR} \
-              --out=${DIFF_OUTPUT_DIR} 2>&1 | tee compare_output.txt
+  # validate-docs:
+  #   name: validate-docs
+  #   runs-on: 'ubuntu-24.04-4core'
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - id: get_commit_msg
+  #       uses: ./.github/actions/get-commit-msg
+  #     - name: Check for manual edits to /docs
+  #       run: |
+  #         COMMIT_ID=$(echo "${{ steps.get_commit_msg.outputs.msg }}" | head -n 1 | sed "s/commit //g")
+  #         bash docs_src/build/presubmit_check.sh ${COMMIT_ID}
 
-          if grep "Failed" compare_output.txt > /dev/null; then
-             DELIMITER="EOF_FILE_CONTENT_$(date +%s)" # Using timestamp to make it more unique
-             echo "err<<$DELIMITER" >> "$GITHUB_OUTPUT"
-             cat compare_output.txt >> "$GITHUB_OUTPUT"
-             echo "$DELIMITER" >> "$GITHUB_OUTPUT"
-          fi
-      - uses: actions/upload-artifact@v4
-        with:
-          name: presubmit-renderdiff-result
-          path: ./out/renderdiff
-      - name: Compare result
-        run: |
-          ERROR_STR="${{ steps.render_compare.outputs.err }}"
-          if [ -n "${ERROR_STR}" ]; then
-              echo "${ERROR_STR}"
-              exit 1
-          fi
+  # test-renderdiff:
+  #   name: test-renderdiff
+  #   runs-on: macos-14-xlarge
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - id: get_commit_msg
+  #       uses: ./.github/actions/get-commit-msg
+  #     - uses: ./.github/actions/mac-prereq
+  #     - uses: ./.github/actions/get-mesa
+  #     - name: Prerequisites
+  #       run: |
+  #         pip install tifffile numpy
+  #       shell: bash
+  #     - name: Render and compare
+  #       id: render_compare
+  #       run: |
+  #         TEST_DIR=test/renderdiff
+  #         source ${TEST_DIR}/src/preamble.sh
+  #         start_
+  #         GOLDEN_BRANCH=$(echo "${{ steps.get_commit_msg.outputs.msg }}" | python3 ${TEST_DIR}/src/commit_msg.py)
+  #         bash ${TEST_DIR}/generate.sh && \
+  #             python3 ${TEST_DIR}/src/golden_manager.py \
+  #                 --branch=${GOLDEN_BRANCH} \
+  #                 --output=${GOLDEN_OUTPUT_DIR}
 
-  validate-wgsl-webgpu:
-    name: validate-wgsl-webgpu
-    runs-on: 'ubuntu-24.04-8core'
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/linux-prereq
-      - name: Run build script
-        run: ./build.sh -W debug test_filamat filament
-      - name: Run test
-        run: ./out/cmake-debug/libs/filamat/test_filamat --gtest_filter=MaterialCompiler.Wgsl*
+  #         # Note that we need to upload the output even if comparison fails, so we undo `set -ex`
+  #         end_
 
-  test-code-correctness:
-    name: test-code-correctness
-    runs-on: 'macos-14-xlarge'
-    steps:
-      - uses: actions/checkout@v4.1.6
-        with:
-          fetch-depth: 0
-      - uses: ./.github/actions/mac-prereq
-      - name: Install clang-tidy and deps
-        run: |
-          pip install pyyaml
-          brew install llvm@${GITHUB_LLVM_VERSION}
-          sudo ln -s "$(brew --prefix llvm)@${GITHUB_LLVM_VERSION}/bin/clang-tidy" "/usr/local/bin/clang-tidy"
-      - name: Run build script
-        # We need to build before clang-tidy can run analysis
-        run: |
-          # This will build for all three desktop backends on mac
-          ./build.sh -p desktop debug gltf_viewer
-      - name: Run test
-        run: bash test/code-correctness/test.sh
+  #         python3 ${TEST_DIR}/src/compare.py \
+  #             --src=${GOLDEN_OUTPUT_DIR} \
+  #             --dest=${RENDER_OUTPUT_DIR} \
+  #             --out=${DIFF_OUTPUT_DIR} 2>&1 | tee compare_output.txt
+
+  #         if grep "Failed" compare_output.txt > /dev/null; then
+  #            DELIMITER="EOF_FILE_CONTENT_$(date +%s)" # Using timestamp to make it more unique
+  #            echo "err<<$DELIMITER" >> "$GITHUB_OUTPUT"
+  #            cat compare_output.txt >> "$GITHUB_OUTPUT"
+  #            echo "$DELIMITER" >> "$GITHUB_OUTPUT"
+  #         fi
+  #     - uses: actions/upload-artifact@v4
+  #       with:
+  #         name: presubmit-renderdiff-result
+  #         path: ./out/renderdiff
+  #     - name: Compare result
+  #       run: |
+  #         ERROR_STR="${{ steps.render_compare.outputs.err }}"
+  #         if [ -n "${ERROR_STR}" ]; then
+  #             echo "${ERROR_STR}"
+  #             exit 1
+  #         fi
+
+  # validate-wgsl-webgpu:
+  #   name: validate-wgsl-webgpu
+  #   runs-on: 'ubuntu-24.04-8core'
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/linux-prereq
+  #     - name: Run build script
+  #       run: ./build.sh -W debug test_filamat filament
+  #     - name: Run test
+  #       run: ./out/cmake-debug/libs/filamat/test_filamat --gtest_filter=MaterialCompiler.Wgsl*
+
+  # test-code-correctness:
+  #   name: test-code-correctness
+  #   runs-on: 'macos-14-xlarge'
+  #   steps:
+  #     - uses: actions/checkout@v4.1.6
+  #       with:
+  #         fetch-depth: 0
+  #     - uses: ./.github/actions/mac-prereq
+  #     - name: Install clang-tidy and deps
+  #       run: |
+  #         pip install pyyaml
+  #         brew install llvm@${GITHUB_LLVM_VERSION}
+  #         sudo ln -s "$(brew --prefix llvm)@${GITHUB_LLVM_VERSION}/bin/clang-tidy" "/usr/local/bin/clang-tidy"
+  #     - name: Run build script
+  #       # We need to build before clang-tidy can run analysis
+  #       run: |
+  #         # This will build for all three desktop backends on mac
+  #         ./build.sh -p desktop debug gltf_viewer
+  #     - name: Run test
+  #       run: bash test/code-correctness/test.sh

CMakeLists.txt

@@ -544,6 +544,7 @@ if (FILAMENT_SUPPORTS_VULKAN)
 endif()
 
 if (FILAMENT_SUPPORTS_WEBGPU)
+    message(AUTHOR_WARNING "WebGPU is in development stage and broken at this stage. Its support is very limited.")
     add_definitions(-DFILAMENT_SUPPORTS_WEBGPU)
 endif()
 
@@ -633,7 +634,7 @@ endif()
 # With WebGPU, Tint does not support ClipDistance which is used in Stereo.  Mentioned in comment
 # https://github.com/google/dawn/blob/855d17b08abdf02f9142bf5a8f14d0ea088810a4/src/tint/lang/spirv/reader/ast_parser/function.cc#L4434
 if (FILAMENT_SUPPORTS_WEBGPU)
-    set(MATC_API_FLAGS ${MATC_API_FLAGS} -a webgpu --variant-filter=stereo)
+    set(MATC_API_FLAGS ${MATC_API_FLAGS} -a webgpu --variant-filter=skinning,stereo)
 endif()
 
 # Disable ESSL 1.0 code generation.

NEW_RELEASE_NOTES.md

@@ -8,4 +8,5 @@ appropriate header in [RELEASE_NOTES.md](./RELEASE_NOTES.md).
 
 ## Release notes for next branch cut
 
-- materials: added `getEyeFromViewMatrix()` for vertex shader [⚠️ **Recompile Materials**]
+- Update CMake minimum version to 3.22.1
+- material: Add a material parameter to control shadow far attenuation (b/436680157)

README.md

@@ -18,7 +18,7 @@ Filament release archives contains host-side tools that are required to generate
 Make sure you always use tools from the same release as the runtime library. This is particularly
 important for `matc` (material compiler).
 
-If you'd rather build Filament yourself, please refer to our [build manual](/BUILDING.md).
+If you'd rather build Filament yourself, please refer to our [build manual](BUILDING.md).
 
 ### Android
 
@@ -31,7 +31,7 @@ repositories {
 }
 
 dependencies {
-    implementation 'com.google.android.filament:filament-android:1.65.0'
+    implementation 'com.google.android.filament:filament-android:1.63.1'
 }
 ```
 
@@ -51,7 +51,7 @@ Here are all the libraries available in the group `com.google.android.filament`:
 iOS projects can use CocoaPods to install the latest release:
 
 ```shell
-pod 'Filament', '~> 1.65.0'
+pod 'Filament', '~> 1.63.1'
 ```
 
 ## Documentation
@@ -222,7 +222,7 @@ MaterialInstance* materialInstance = material->createInstance();
 ```
 
 To learn more about materials and `matc`, please refer to the
-[materials documentation](https://google.github.io/filament/Materials.html).
+[materials documentation](./docs/Materials.md.html).
 
 To render, simply pass the `View` to the `Renderer`:
 
@@ -240,7 +240,7 @@ in the `samples/` directory. These samples are all based on `libs/filamentapp/`
 code that creates a native window with SDL2 and initializes the Filament engine, renderer and views.
 
 For more information on how to prepare environment maps for image-based lighting please refer to
-[BUILDING.md](/BUILDING.md#running-the-native-samples).
+[BUILDING.md](https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples).
 
 ### Android
 
@@ -272,7 +272,7 @@ To get started you can use the textures and environment maps found respectively
 refer to their respective `URL.txt` files to know more about the original authors.
 
 Environments must be pre-processed using
-[`cmgen`](/BUILDING.md#running-the-native-samples) or
+[`cmgen`](https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples) or
 using the `libiblprefilter` library.
 
 ## How to make contributions

RELEASE_NOTES.md

@@ -7,15 +7,6 @@ A new header is inserted each time a *tag* is created.
 Instead, if you are authoring a PR for the main branch, add your release note to
 [NEW_RELEASE_NOTES.md](./NEW_RELEASE_NOTES.md).
 
-## v1.65.0
-
-lighting: the intermediate froxel record buffer is now dynamically sized [⚠️ **New Material Version**]
-
-## v1.64.1
-
-- Update CMake minimum version to 3.22.1
-- material: Add a material parameter to control shadow far attenuation (b/436680157)
-
 ## v1.64.0
 
 - engine: add a `linearFog` material parameter. [⚠️ **New Material Version**]

android/buildSrc/src/main/groovy/FilamentPlugin.groovy

@@ -146,7 +146,7 @@ abstract class MaterialCompiler extends TaskWithBinary {
                         .gradleProperty("com.google.android.filament.include-webgpu")
                         .forUseAtConfigurationTime().present
                 if (include_webgpu) {
-                    matcArgs += ['-a', 'webgpu', '--variant-filter=stereo']
+                    matcArgs += ['-a', 'webgpu', '--variant-filter=skinning,stereo']
                 }
 
                 def mat_no_opt = providers

android/filament-android/src/main/java/com/google/android/filament/View.java

@@ -1949,18 +1949,6 @@ public class View {
          * Ground Truth-base Ambient Occlusion (GTAO) options
          */
         public float gtaoThicknessHeuristic = 0.004f;
-        /**
-         * Ground Truth-base Ambient Occlusion (GTAO) options
-         */
-        public boolean gtaoUseVisibilityBitmasks = false;
-        /**
-         * Ground Truth-base Ambient Occlusion (GTAO) options
-         */
-        public float gtaoConstThickness = 0.5f;
-        /**
-         * Ground Truth-base Ambient Occlusion (GTAO) options
-         */
-        public boolean gtaoLinearThickness = false;
 
     }
 

android/gradle.properties

@@ -1,5 +1,5 @@
 GROUP=com.google.android.filament
-VERSION_NAME=1.65.0
+VERSION_NAME=1.63.1
 
 POM_DESCRIPTION=Real-time physically based rendering engine for Android.
 

build/common/get-gltf-sample-assets.sh

@@ -1,60 +0,0 @@
-# Copyright (C) 2025 The Android Open Source Project
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-#!/usr/bin/bash
-set -e
-
-GLTF_SAMPLE_ASSETS_COMMIT=${GITHUB_GLTF_SAMPLE_ASSETS_COMMIT:-d441dfdb87413ff412c620849a649d61789a470f}
-COMMIT_HASH="${GLTF_SAMPLE_ASSETS_COMMIT}"
-REPO_URL="https://github.com/KhronosGroup/glTF-Sample-Assets.git"
-TARGET_DIR="gltf"
-
-# The default directories to check out if none are specified
-DEFAULT_SPARSE_PATHS=(
-    "Models/Box/"
-    "Models/Triangle/"
-    "Models/AnimatedCube/"
-)
-
-# Check if command-line arguments are provided
-if [ "$#" -gt 0 ]; then
-    # If arguments are provided, use them as the paths
-    SPARSE_PATHS=()
-    for model_name in "$@"; do
-        SPARSE_PATHS+=("Models/${model_name}/")
-    done
-    echo "Downloading specified models: $@"
-else
-    # Otherwise, use the default list
-    SPARSE_PATHS=("${DEFAULT_SPARSE_PATHS[@]}")
-    echo "No models specified, downloading default set."
-fi
-
-echo "Removing old directory..."
-rm -rf "${TARGET_DIR}"
-
-# Clone the repository using a "treeless" clone, which is highly efficient.
-# --filter=tree:0: Clones only the repository structure without file content (no historical directory listings), making the initial clone very small.
-# --no-checkout:   Prevents automatically checking out the main branch. We will check out a specific commit later.
-# --sparse:        Initializes the repository for sparse checkout, allowing us to fetch only specific directories.
-git clone --filter=tree:0 --no-checkout --sparse "${REPO_URL}" "${TARGET_DIR}"
-
-cd "${TARGET_DIR}"
-
-git sparse-checkout set "${SPARSE_PATHS[@]}"
-
-echo "Checking out commit ${COMMIT_HASH}..."
-git checkout "${COMMIT_HASH}"
-
-echo "Successfully checked out the specified models into the '${TARGET_DIR}' directory."

build/common/get-mesa.sh

@@ -137,9 +137,6 @@ if  [[ "$OS_NAME" == "Darwin" ]]; then
 
     # This is necessary to be able to build vk (lavapipe) on macOS.  Doesn't seem like a real dependency.
     sed -I '' "s/error('Vulkan drivers require dri3 for X11 support')//g" meson.build
-    # This is to properly link lib-xcb-present on the mac build (though we won't be drawing to any
-    # real hardware surface).
-    sed -I '' "s/dep_xcb_present = null_dep/dep_xcb_present = dependency('xcb-present')/g" meson.build
 fi
 
 # -Dosmesa=true    => builds OSMesa, which is an offscreen GL context

build/common/get-vulkan-sdk.sh

@@ -109,8 +109,13 @@ function install_mac() {
     unzip -t vulkan_sdk.zip
   fi
   echo "recognized zip layout 'vulkan_sdk.zip' ${InstallVulkan}.app/Contents" >&2
-  ${InstallVulkan}.app/Contents/MacOS/${InstallVulkan} --accept-licenses --default-answer --confirm-command install
+  local sdk_temp=${VULKAN_SDK_DIR}.tmp
+  sudo ${InstallVulkan}.app/Contents/MacOS/${InstallVulkan} --root "$sdk_temp" --accept-licenses --default-answer --confirm-command install
+  du -hs $sdk_temp
+  test -d $sdk_temp/macOS || { echo "unrecognized dmg folder layout: $sdk_temp" ; ls -l $sdk_temp ; }
+  cp -r $sdk_temp/macOS/* $VULKAN_SDK_DIR/
   if [[ -d ${InstallVulkan}.app/Contents ]] ; then
+    sudo rm -rf "$sdk_temp"
     rm -rf ${InstallVulkan}.app
   fi
 }

build/common/versions

@@ -5,5 +5,4 @@ GITHUB_MESA_VERSION=24.2.1
 GITHUB_LLVM_VERSION=16
 GITHUB_NDK_VERSION=27.0.11718014
 GITHUB_EMSDK_VERSION=3.1.60
-GITHUB_VULKANSDK_VERSION=1.4.321.0
-GITHUB_GLTF_SAMPLE_ASSETS_COMMIT=d441dfdb87413ff412c620849a649d61789a470f
+GITHUB_VULKANSDK_VERSION=1.4.321.0

build/windows/build-github.bat

@@ -135,7 +135,9 @@ set build_flags=-j %NUMBER_OF_PROCESSORS%
 :: cmake --build . --target sample-resources --config %config% %build_flags% || exit /b
 :: cmake --build . --target suzanne-resources --config %config% %build_flags% || exit /b
 
-cmake --build . %INSTALL% --config %config% %build_flags% -- /m || exit /b
+cmake --build . --target sample-resources --config %config% %build_flags% -- /m || exit /b
+
+:: cmake --build . %INSTALL% --config %config% %build_flags% -- /m || exit /b
 @echo off
 
 echo Disk info after building variant: %variant%

docs/dup/docs.html

@@ -221,8 +221,6 @@ export is an html with the output captured in a <code>&lt;pre&gt;</code> tag).</
 <li>Replace css styling in the exported output as needed (so they don't interfere with the book's css.</li>
 <li>Replace resource urls to refer to locations relative to the mdbook structure.</li>
 </ul>
-<p>Any <code>markdeep</code> doc can be placed in <code>docs_src/src_markdeep/</code> and they will be parsed to html and included
-in the book as above.</p>
 <h3 id="readmes"><a class="header" href="#readmes">READMEs</a></h3>
 <p>Filament depends on a number of libraries, which reside in the directory <code>libs</code>. These individual
 libaries often have README.md in their root to describe itself. We collect these descriptions into our
@@ -231,7 +229,7 @@ located in <code>tools</code>. Some of tools also have README.md as description.
 <p>The process for copying and processing these READMEs is outlined in <a href="#introductory-doc">Introductory docs</a>.</p>
 <h3 id="other-technical-notes"><a class="header" href="#other-technical-notes">Other technical notes</a></h3>
 <p>These are technical documents that do not fit into a library, tool, or directory of the
-Filament source tree. We collect them into the <code>docs_src/src_mdbook/src/notes</code> directory. No additional
+Filament source tree. We collect them into the <code>docs_src/src/notes</code> directory. No additional
 processing is needed for these documents.</p>
 <h3 id="raw-source-files"><a class="header" href="#raw-source-files">Raw source files</a></h3>
 <p>These are files that are not part of the <code>mdbook</code> generation, but should be included output in <code>/docs</code>
@@ -244,7 +242,7 @@ add a link in <code>SUMMARY.md</code>, and perform the steps outlined in
 <a href="#how-to-create">how-to create section</a>.</p>
 <p>For example, if you are adding a general technical note, then you would</p>
 <ul>
-<li>Place the document (file with extension <code>.md</code>) in <code>docs_src/src_mdbook/src/notes</code></li>
+<li>Place the document (file with extension <code>.md</code>) in <code>docs_src/src/notes</code></li>
 <li>Add a link in <a href="https://github.com/google/filament/blob/main/docs_src/src_mdbook/src/SUMMARY.md"><code>docs_src/src_mdbook/src/SUMMARY.md</code></a></li>
 <li>Run the commands in the <a href="#how-to-generate">Generate</a> section</li>
 </ul>

docs/dup/intro.html

@@ -181,7 +181,7 @@ important for <code>matc</code> (material compiler).</p>
 }
 
 dependencies {
-    implementation 'com.google.android.filament:filament-android:1.64.0'
+    implementation 'com.google.android.filament:filament-android:1.63.1'
 }
 </code></pre>
 <p>Here are all the libraries available in the group <code>com.google.android.filament</code>:</p>
@@ -196,18 +196,18 @@ dependencies {
 </div>
 <h3 id="ios"><a class="header" href="#ios">iOS</a></h3>
 <p>iOS projects can use CocoaPods to install the latest release:</p>
-<pre><code class="language-shell">pod 'Filament', '~&gt; 1.64.0'
+<pre><code class="language-shell">pod 'Filament', '~&gt; 1.63.1'
 </code></pre>
 <h2 id="documentation"><a class="header" href="#documentation">Documentation</a></h2>
 <ul>
-<li><a href="../main/filament.html">Filament</a>, an in-depth explanation of
+<li><a href="https://google.github.io/filament/Filament.html">Filament</a>, an in-depth explanation of
 real-time physically based rendering, the graphics capabilities and implementation of Filament.
 This document explains the math and reasoning behind most of our decisions. This document is a
 good introduction to PBR for graphics programmers.</li>
-<li><a href="../main/materials.html">Materials</a>, the full reference
+<li><a href="https://google.github.io/filament/Materials.html">Materials</a>, the full reference
 documentation for our material system. This document explains our different material models, how
 to use the material compiler <code>matc</code> and how to write custom materials.</li>
-<li><a href="../notes/material_properties.html">Material Properties</a>, a reference
+<li><a href="https://google.github.io/filament/notes/material_properties.html">Material Properties</a>, a reference
 sheet for the standard material model.</li>
 </ul>
 <h2 id="examples"><a class="header" href="#examples">Examples</a></h2>
@@ -395,7 +395,7 @@ by <code>matc</code>:</p>
 MaterialInstance* materialInstance = material-&gt;createInstance();
 </code></pre>
 <p>To learn more about materials and <code>matc</code>, please refer to the
-<a href="../main/materials.html">materials documentation</a>.</p>
+<a href="./docs/Materials.html">materials documentation</a>.</p>
 <p>To render, simply pass the <code>View</code> to the <code>Renderer</code>:</p>
 <pre><code class="language-c++">// beginFrame() returns false if we need to skip a frame
 if (renderer-&gt;beginFrame(swapChain)) {
@@ -408,7 +408,7 @@ if (renderer-&gt;beginFrame(swapChain)) {
 in the <code>samples/</code> directory. These samples are all based on <code>libs/filamentapp/</code> which contains the
 code that creates a native window with SDL2 and initializes the Filament engine, renderer and views.</p>
 <p>For more information on how to prepare environment maps for image-based lighting please refer to
-<a href="building.html#running-the-native-samples">BUILDING.md</a>.</p>
+<a href="https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples">BUILDING.md</a>.</p>
 <h3 id="android-1"><a class="header" href="#android-1">Android</a></h3>
 <p>See <code>android/samples</code> for examples of how to use Filament on Android.</p>
 <p>You must always first initialize Filament by calling <code>Filament.init()</code>.</p>
@@ -430,7 +430,7 @@ OpenGL ES.</p>
 <code>third_party/textures</code> and <code>third_party/environments</code>. These assets are under CC0 license. Please
 refer to their respective <code>URL.txt</code> files to know more about the original authors.</p>
 <p>Environments must be pre-processed using
-<a href="building.html#running-the-native-samples"><code>cmgen</code></a> or
+<a href="https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples"><code>cmgen</code></a> or
 using the <code>libiblprefilter</code> library.</p>
 <h2 id="how-to-make-contributions"><a class="header" href="#how-to-make-contributions">How to make contributions</a></h2>
 <p>Please read and follow the steps in <a href="contributing.html">CONTRIBUTING.md</a>. Make sure you are
@@ -513,7 +513,7 @@ and tools.</p>
 <li><code>web</code>:                      JavaScript bindings, documentation, and samples</li>
 </ul>
 <h2 id="license"><a class="header" href="#license">License</a></h2>
-<p>Please see <a href="https://github.com/google/filament/blob/main/LICENSE">LICENSE</a>.</p>
+<p>Please see <a href="/LICENSE">LICENSE</a>.</p>
 <h2 id="disclaimer"><a class="header" href="#disclaimer">Disclaimer</a></h2>
 <p>This is not an officially supported Google product.</p>
 

docs/index.html

@@ -1,11 +1,564 @@
-<!DOCTYPE html>
-<html lang="en">
-<head>
-    <meta charset="UTF-8">
-    <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Redirecting...</title>
-    <meta http-equiv="refresh" content="0; url=./dup/intro.html">
-</head>
-<body>
-</body>
+<!DOCTYPE HTML>
+<html lang="en" class="light sidebar-visible" dir="ltr">
+    <head>
+        <!-- Book generated using mdBook -->
+        <meta charset="UTF-8">
+        <title>Introduction - Filament</title>
+
+
+        <!-- Custom HTML head -->
+
+        <meta name="description" content="">
+        <meta name="viewport" content="width=device-width, initial-scale=1">
+        <meta name="theme-color" content="#ffffff">
+
+        <link rel="shortcut icon" href="favicon.png">
+        <link rel="stylesheet" href="css/variables.css">
+        <link rel="stylesheet" href="css/general.css">
+        <link rel="stylesheet" href="css/chrome.css">
+
+        <!-- Fonts -->
+        <link rel="stylesheet" href="FontAwesome/css/font-awesome.css">
+        <link rel="stylesheet" href="fonts/fonts.css">
+
+        <!-- Highlight.js Stylesheets -->
+        <link rel="stylesheet" href="highlight.css">
+        <link rel="stylesheet" href="tomorrow-night.css">
+        <link rel="stylesheet" href="ayu-highlight.css">
+
+        <!-- Custom theme stylesheets -->
+
+        <!-- MathJax -->
+        <script async src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
+
+        <!-- Provide site root to javascript -->
+        <script>
+            var path_to_root = "";
+            var default_theme = window.matchMedia("(prefers-color-scheme: dark)").matches ? "light" : "light";
+        </script>
+        <!-- Start loading toc.js asap -->
+        <script src="toc.js"></script>
+    </head>
+    <body>
+    <div id="body-container">
+        <!-- Work around some values being stored in localStorage wrapped in quotes -->
+        <script>
+            try {
+                var theme = localStorage.getItem('mdbook-theme');
+                var sidebar = localStorage.getItem('mdbook-sidebar');
+
+                if (theme.startsWith('"') && theme.endsWith('"')) {
+                    localStorage.setItem('mdbook-theme', theme.slice(1, theme.length - 1));
+                }
+
+                if (sidebar.startsWith('"') && sidebar.endsWith('"')) {
+                    localStorage.setItem('mdbook-sidebar', sidebar.slice(1, sidebar.length - 1));
+                }
+            } catch (e) { }
+        </script>
+
+        <!-- Set the theme before any content is loaded, prevents flash -->
+        <script>
+            var theme;
+            try { theme = localStorage.getItem('mdbook-theme'); } catch(e) { }
+            if (theme === null || theme === undefined) { theme = default_theme; }
+            const html = document.documentElement;
+            html.classList.remove('light')
+            html.classList.add(theme);
+            html.classList.add("js");
+        </script>
+
+        <input type="checkbox" id="sidebar-toggle-anchor" class="hidden">
+
+        <!-- Hide / unhide sidebar before it is displayed -->
+        <script>
+            var sidebar = null;
+            var sidebar_toggle = document.getElementById("sidebar-toggle-anchor");
+            if (document.body.clientWidth >= 1080) {
+                try { sidebar = localStorage.getItem('mdbook-sidebar'); } catch(e) { }
+                sidebar = sidebar || 'visible';
+            } else {
+                sidebar = 'hidden';
+            }
+            sidebar_toggle.checked = sidebar === 'visible';
+            html.classList.remove('sidebar-visible');
+            html.classList.add("sidebar-" + sidebar);
+        </script>
+
+        <nav id="sidebar" class="sidebar" aria-label="Table of contents">
+            <div style="display:flex;align-items:center;justify-content:center">
+                <img class="flogo" src="images/filament_logo_small.png"></img>
+            </div>
+            <!-- populated by js -->
+            <mdbook-sidebar-scrollbox class="sidebar-scrollbox"></mdbook-sidebar-scrollbox>
+            <noscript>
+                <iframe class="sidebar-iframe-outer" src="toc.html"></iframe>
+            </noscript>
+            <div id="sidebar-resize-handle" class="sidebar-resize-handle">
+                <div class="sidebar-resize-indicator"></div>
+            </div>
+        </nav>
+
+        <div id="page-wrapper" class="page-wrapper">
+
+            <div class="page">
+                <div id="menu-bar-hover-placeholder"></div>
+                <div id="menu-bar" class="menu-bar sticky">
+                    <div class="left-buttons">
+                        <label id="sidebar-toggle" class="icon-button" for="sidebar-toggle-anchor" title="Toggle Table of Contents" aria-label="Toggle Table of Contents" aria-controls="sidebar">
+                            <i class="fa fa-bars"></i>
+                        </label>
+                        <!-- Filament: disable themes because the markdeep part does not look good for dark themes -->
+                        <!--
+                        <button id="theme-toggle" class="icon-button" type="button" title="Change theme" aria-label="Change theme" aria-haspopup="true" aria-expanded="false" aria-controls="theme-list">
+                            <i class="fa fa-paint-brush"></i>
+                        </button>
+                        <ul id="theme-list" class="theme-popup" aria-label="Themes" role="menu">
+                            <li role="none"><button role="menuitem" class="theme" id="light">Light</button></li>
+                            <li role="none"><button role="menuitem" class="theme" id="rust">Rust</button></li>
+                            <li role="none"><button role="menuitem" class="theme" id="coal">Coal</button></li>
+                            <li role="none"><button role="menuitem" class="theme" id="navy">Navy</button></li>
+                            <li role="none"><button role="menuitem" class="theme" id="ayu">Ayu</button></li>
+                        </ul>
+                        -->
+                        <button id="search-toggle" class="icon-button" type="button" title="Search. (Shortkey: s)" aria-label="Toggle Searchbar" aria-expanded="false" aria-keyshortcuts="S" aria-controls="searchbar">
+                            <i class="fa fa-search"></i>
+                        </button>
+                    </div>
+
+                    <h1 class="menu-title">Filament</h1>
+
+                    <div class="right-buttons">
+                        <a href="https://github.com/google/filament" title="Git repository" aria-label="Git repository">
+                            <i id="git-repository-button" class="fa fa-github"></i>
+                        </a>
+
+                    </div>
+                </div>
+
+                <div id="search-wrapper" class="hidden">
+                    <form id="searchbar-outer" class="searchbar-outer">
+                        <input type="search" id="searchbar" name="searchbar" placeholder="Search this book ..." aria-controls="searchresults-outer" aria-describedby="searchresults-header">
+                    </form>
+                    <div id="searchresults-outer" class="searchresults-outer hidden">
+                        <div id="searchresults-header" class="searchresults-header"></div>
+                        <ul id="searchresults">
+                        </ul>
+                    </div>
+                </div>
+
+                <!-- Apply ARIA attributes after the sidebar and the sidebar toggle button are added to the DOM -->
+                <script>
+                    document.getElementById('sidebar-toggle').setAttribute('aria-expanded', sidebar === 'visible');
+                    document.getElementById('sidebar').setAttribute('aria-hidden', sidebar !== 'visible');
+                    Array.from(document.querySelectorAll('#sidebar a')).forEach(function(link) {
+                        link.setAttribute('tabIndex', sidebar === 'visible' ? 0 : -1);
+                    });
+                </script>
+
+                <div id="content" class="content">
+                    <main>
+                        <h1 id="filament"><a class="header" href="#filament">Filament</a></h1>
+<p><a href="https://github.com/google/filament/actions?query=workflow%3AAndroid"><img src="https://github.com/google/filament/workflows/Android/badge.svg" alt="Android Build Status" /></a>
+<a href="https://github.com/google/filament/actions?query=workflow%3AiOS"><img src="https://github.com/google/filament/workflows/iOS/badge.svg" alt="iOS Build Status" /></a>
+<a href="https://github.com/google/filament/actions?query=workflow%3ALinux"><img src="https://github.com/google/filament/workflows/Linux/badge.svg" alt="Linux Build Status" /></a>
+<a href="https://github.com/google/filament/actions?query=workflow%3AmacOS"><img src="https://github.com/google/filament/workflows/macOS/badge.svg" alt="macOS Build Status" /></a>
+<a href="https://github.com/google/filament/actions?query=workflow%3AWindows"><img src="https://github.com/google/filament/workflows/Windows/badge.svg" alt="Windows Build Status" /></a>
+<a href="https://github.com/google/filament/actions?query=workflow%3AWeb"><img src="https://github.com/google/filament/workflows/Web/badge.svg" alt="Web Build Status" /></a></p>
+<p>Filament is a real-time physically based rendering engine for Android, iOS, Linux, macOS, Windows,
+and WebGL. It is designed to be as small as possible and as efficient as possible on Android.</p>
+<h2 id="download"><a class="header" href="#download">Download</a></h2>
+<p><a href="https://github.com/google/filament/releases">Download Filament releases</a> to access stable builds.
+Filament release archives contains host-side tools that are required to generate assets.</p>
+<p>Make sure you always use tools from the same release as the runtime library. This is particularly
+important for <code>matc</code> (material compiler).</p>
+<p>If you'd rather build Filament yourself, please refer to our <a href="building.html">build manual</a>.</p>
+<h3 id="android"><a class="header" href="#android">Android</a></h3>
+<p>Android projects can simply declare Filament libraries as Maven dependencies:</p>
+<pre><code class="language-gradle">repositories {
+    // ...
+    mavenCentral()
+}
+
+dependencies {
+    implementation 'com.google.android.filament:filament-android:1.63.1'
+}
+</code></pre>
+<p>Here are all the libraries available in the group <code>com.google.android.filament</code>:</p>
+<div class="table-wrapper"><table><thead><tr><th>Artifact</th><th>Description</th></tr></thead><tbody>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-android"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-android/badge.svg?subject=filament-android" alt="filament-android" /></a></td><td>The Filament rendering engine itself.</td></tr>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-android-debug"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-android-debug/badge.svg?subject=filament-android-debug" alt="filament-android-debug" /></a></td><td>Debug version of <code>filament-android</code>.</td></tr>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/gltfio-android"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/gltfio-android/badge.svg?subject=gltfio-android" alt="gltfio-android" /></a></td><td>A glTF 2.0 loader for Filament, depends on <code>filament-android</code>.</td></tr>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-utils-android"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filament-utils-android/badge.svg?subject=filament-utils-android" alt="filament-utils-android" /></a></td><td>KTX loading, Kotlin math, and camera utilities, depends on <code>gltfio-android</code>.</td></tr>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filamat-android"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filamat-android/badge.svg?subject=filamat-android" alt="filamat-android" /></a></td><td>A runtime material builder/compiler. This library is large but contains a full shader compiler/validator/optimizer and supports both OpenGL and Vulkan.</td></tr>
+<tr><td><a href="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filamat-android-lite"><img src="https://maven-badges.herokuapp.com/maven-central/com.google.android.filament/filamat-android-lite/badge.svg?subject=filamat-android-lite" alt="filamat-android-lite" /></a></td><td>A much smaller alternative to <code>filamat-android</code> that can only generate OpenGL shaders. It does not provide validation or optimizations.</td></tr>
+</tbody></table>
+</div>
+<h3 id="ios"><a class="header" href="#ios">iOS</a></h3>
+<p>iOS projects can use CocoaPods to install the latest release:</p>
+<pre><code class="language-shell">pod 'Filament', '~&gt; 1.63.1'
+</code></pre>
+<h2 id="documentation"><a class="header" href="#documentation">Documentation</a></h2>
+<ul>
+<li><a href="https://google.github.io/filament/Filament.html">Filament</a>, an in-depth explanation of
+real-time physically based rendering, the graphics capabilities and implementation of Filament.
+This document explains the math and reasoning behind most of our decisions. This document is a
+good introduction to PBR for graphics programmers.</li>
+<li><a href="https://google.github.io/filament/Materials.html">Materials</a>, the full reference
+documentation for our material system. This document explains our different material models, how
+to use the material compiler <code>matc</code> and how to write custom materials.</li>
+<li><a href="https://google.github.io/filament/notes/material_properties.html">Material Properties</a>, a reference
+sheet for the standard material model.</li>
+</ul>
+<h2 id="examples"><a class="header" href="#examples">Examples</a></h2>
+<p><img src="../images/samples/example_bistro1.jpg" alt="Night scene" />
+<img src="../images/samples/example_bistro2.jpg" alt="Night scene" />
+<img src="../images/samples/example_materials1.jpg" alt="Materials" />
+<img src="../images/samples/example_materials2.jpg" alt="Materials" />
+<img src="../images/samples/example_helmet.jpg" alt="Helmet" />
+<img src="../images/samples/example_ssr.jpg" alt="Screen-space refraction" /></p>
+<h2 id="features"><a class="header" href="#features">Features</a></h2>
+<h3 id="apis"><a class="header" href="#apis">APIs</a></h3>
+<ul>
+<li>Native C++ API for Android, iOS, Linux, macOS and Windows</li>
+<li>Java/JNI API for Android</li>
+<li>JavaScript API</li>
+</ul>
+<h3 id="backends"><a class="header" href="#backends">Backends</a></h3>
+<ul>
+<li>OpenGL 4.1+ for Linux, macOS and Windows</li>
+<li>OpenGL ES 3.0+ for Android and iOS</li>
+<li>Metal for macOS and iOS</li>
+<li>Vulkan 1.0 for Android, Linux, macOS, and Windows</li>
+<li>WebGL 2.0 for all platforms</li>
+</ul>
+<h3 id="rendering"><a class="header" href="#rendering">Rendering</a></h3>
+<ul>
+<li>Clustered forward renderer</li>
+<li>Cook-Torrance microfacet specular BRDF</li>
+<li>Lambertian diffuse BRDF</li>
+<li>Custom lighting/surface shading</li>
+<li>HDR/linear lighting</li>
+<li>Metallic workflow</li>
+<li>Clear coat</li>
+<li>Anisotropic lighting</li>
+<li>Approximated translucent (subsurface) materials</li>
+<li>Cloth/fabric/sheen shading</li>
+<li>Normal mapping &amp; ambient occlusion mapping</li>
+<li>Image-based lighting</li>
+<li>Physically-based camera (shutter speed, sensitivity and aperture)</li>
+<li>Physical light units</li>
+<li>Point lights, spot lights, and directional light</li>
+<li>Specular anti-aliasing</li>
+<li>Point, spot, and directional light shadows</li>
+<li>Cascaded shadows</li>
+<li>EVSM, PCSS, DPCF, or PCF shadows</li>
+<li>Transparent shadows</li>
+<li>Contact shadows</li>
+<li>Screen-space ambient occlusion</li>
+<li>Screen-space reflections</li>
+<li>Screen-space refraction</li>
+<li>Global fog</li>
+<li>Dynamic resolution (with support for AMD FidelityFX FSR)</li>
+</ul>
+<h3 id="post-processing"><a class="header" href="#post-processing">Post processing</a></h3>
+<ul>
+<li>HDR bloom</li>
+<li>Depth of field bokeh</li>
+<li>Multiple tone mappers: generic (customizable), ACES, filmic, etc.</li>
+<li>Color and tone management: luminance scaling, gamut mapping</li>
+<li>Color grading: exposure, night adaptation, white balance, channel mixer,
+shadows/mid-tones/highlights, ASC CDL, contrast, saturation, etc.</li>
+<li>TAA, FXAA, MSAA</li>
+<li>Screen-space lens flares</li>
+</ul>
+<h3 id="gltf-20"><a class="header" href="#gltf-20">glTF 2.0</a></h3>
+<ul>
+<li>
+<p>Encodings</p>
+<ul>
+<li><input disabled="" type="checkbox" checked=""/>
+Embeded</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Binary</li>
+</ul>
+</li>
+<li>
+<p>Primitive Types</p>
+<ul>
+<li><input disabled="" type="checkbox" checked=""/>
+Points</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Lines</li>
+<li><input disabled="" type="checkbox"/>
+Line Loop</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Line Strip</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Triangles</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Triangle Strip</li>
+<li><input disabled="" type="checkbox"/>
+Triangle Fan</li>
+</ul>
+</li>
+<li>
+<p>Animation</p>
+<ul>
+<li><input disabled="" type="checkbox" checked=""/>
+Transform animation</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Linear interpolation</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Morph animation
+<ul>
+<li><input disabled="" type="checkbox" checked=""/>
+Sparse accessor</li>
+</ul>
+</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Skin animation</li>
+<li><input disabled="" type="checkbox" checked=""/>
+Joint animation</li>
+</ul>
+</li>
+<li>
+<p>Extensions</p>
+<ul>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_draco_mesh_compression</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_lights_punctual</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_clearcoat</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_emissive_strength</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_ior</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_pbrSpecularGlossiness</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_sheen</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_transmission</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_unlit</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_variants</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_volume</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_materials_specular</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_mesh_quantization</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_texture_basisu</li>
+<li><input disabled="" type="checkbox" checked=""/>
+KHR_texture_transform</li>
+<li><input disabled="" type="checkbox" checked=""/>
+EXT_meshopt_compression</li>
+</ul>
+</li>
+</ul>
+<h2 id="rendering-with-filament"><a class="header" href="#rendering-with-filament">Rendering with Filament</a></h2>
+<h3 id="native-linux-macos-and-windows"><a class="header" href="#native-linux-macos-and-windows">Native Linux, macOS and Windows</a></h3>
+<p>You must create an <code>Engine</code>, a <code>Renderer</code> and a <code>SwapChain</code>. The <code>SwapChain</code> is created from a
+native window pointer (an <code>NSView</code> on macOS or a <code>HWND</code> on Windows for instance):</p>
+<pre><code class="language-c++">Engine* engine = Engine::create();
+SwapChain* swapChain = engine-&gt;createSwapChain(nativeWindow);
+Renderer* renderer = engine-&gt;createRenderer();
+</code></pre>
+<p>To render a frame you must then create a <code>View</code>, a <code>Scene</code> and a <code>Camera</code>:</p>
+<pre><code class="language-c++">Camera* camera = engine-&gt;createCamera(EntityManager::get().create());
+View* view = engine-&gt;createView();
+Scene* scene = engine-&gt;createScene();
+
+view-&gt;setCamera(camera);
+view-&gt;setScene(scene);
+</code></pre>
+<p>Renderables are added to the scene:</p>
+<pre><code class="language-c++">Entity renderable = EntityManager::get().create();
+// build a quad
+RenderableManager::Builder(1)
+        .boundingBox({{ -1, -1, -1 }, { 1, 1, 1 }})
+        .material(0, materialInstance)
+        .geometry(0, RenderableManager::PrimitiveType::TRIANGLES, vertexBuffer, indexBuffer, 0, 6)
+        .culling(false)
+        .build(*engine, renderable);
+scene-&gt;addEntity(renderable);
+</code></pre>
+<p>The material instance is obtained from a material, itself loaded from a binary blob generated
+by <code>matc</code>:</p>
+<pre><code class="language-c++">Material* material = Material::Builder()
+        .package((void*) BAKED_MATERIAL_PACKAGE, sizeof(BAKED_MATERIAL_PACKAGE))
+        .build(*engine);
+MaterialInstance* materialInstance = material-&gt;createInstance();
+</code></pre>
+<p>To learn more about materials and <code>matc</code>, please refer to the
+<a href="./docs/Materials.html">materials documentation</a>.</p>
+<p>To render, simply pass the <code>View</code> to the <code>Renderer</code>:</p>
+<pre><code class="language-c++">// beginFrame() returns false if we need to skip a frame
+if (renderer-&gt;beginFrame(swapChain)) {
+    // for each View
+    renderer-&gt;render(view);
+    renderer-&gt;endFrame();
+}
+</code></pre>
+<p>For complete examples of Linux, macOS and Windows Filament applications, look at the source files
+in the <code>samples/</code> directory. These samples are all based on <code>libs/filamentapp/</code> which contains the
+code that creates a native window with SDL2 and initializes the Filament engine, renderer and views.</p>
+<p>For more information on how to prepare environment maps for image-based lighting please refer to
+<a href="https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples">BUILDING.md</a>.</p>
+<h3 id="android-1"><a class="header" href="#android-1">Android</a></h3>
+<p>See <code>android/samples</code> for examples of how to use Filament on Android.</p>
+<p>You must always first initialize Filament by calling <code>Filament.init()</code>.</p>
+<p>Rendering with Filament on Android is similar to rendering from native code (the APIs are largely
+the same across languages). You can render into a <code>Surface</code> by passing a <code>Surface</code> to the
+<code>createSwapChain</code> method. This allows you to render to a <code>SurfaceTexture</code>, a <code>TextureView</code> or
+a <code>SurfaceView</code>. To make things easier we provide an Android specific API called <code>UiHelper</code> in the
+package <code>com.google.android.filament.android</code>. All you need to do is set a render callback on the
+helper and attach your <code>SurfaceView</code> or <code>TextureView</code> to it. You are still responsible for
+creating the swap chain in the <code>onNativeWindowChanged()</code> callback.</p>
+<h3 id="ios-1"><a class="header" href="#ios-1">iOS</a></h3>
+<p>Filament is supported on iOS 11.0 and above. See <code>ios/samples</code> for examples of using Filament on
+iOS.</p>
+<p>Filament on iOS is largely the same as native rendering with C++. A <code>CAEAGLLayer</code> or <code>CAMetalLayer</code>
+is passed to the <code>createSwapChain</code> method. Filament for iOS supports both Metal (preferred) and
+OpenGL ES.</p>
+<h2 id="assets"><a class="header" href="#assets">Assets</a></h2>
+<p>To get started you can use the textures and environment maps found respectively in
+<code>third_party/textures</code> and <code>third_party/environments</code>. These assets are under CC0 license. Please
+refer to their respective <code>URL.txt</code> files to know more about the original authors.</p>
+<p>Environments must be pre-processed using
+<a href="https://github.com/google/filament/blob/main/BUILDING.md#running-the-native-samples"><code>cmgen</code></a> or
+using the <code>libiblprefilter</code> library.</p>
+<h2 id="how-to-make-contributions"><a class="header" href="#how-to-make-contributions">How to make contributions</a></h2>
+<p>Please read and follow the steps in <a href="contributing.html">CONTRIBUTING.md</a>. Make sure you are
+familiar with the <a href="code_style.html">code style</a>.</p>
+<h2 id="directory-structure"><a class="header" href="#directory-structure">Directory structure</a></h2>
+<p>This repository not only contains the core Filament engine, but also its supporting libraries
+and tools.</p>
+<ul>
+<li><code>android</code>:                  Android libraries and projects
+<ul>
+<li><code>filamat-android</code>:        Filament material generation library (AAR) for Android</li>
+<li><code>filament-android</code>:       Filament library (AAR) for Android</li>
+<li><code>filament-utils-android</code>: Extra utilities (KTX loader, math types, etc.)</li>
+<li><code>gltfio-android</code>:         Filament glTF loading library (AAR) for Android</li>
+<li><code>samples</code>:                Android-specific Filament samples</li>
+</ul>
+</li>
+<li><code>art</code>:                      Source for various artworks (logos, PDF manuals, etc.)</li>
+<li><code>assets</code>:                   3D assets to use with sample applications</li>
+<li><code>build</code>:                    CMake build scripts</li>
+<li><code>docs</code>:                     Documentation
+<ul>
+<li><code>math</code>:                   Mathematica notebooks used to explore BRDFs, equations, etc.</li>
+</ul>
+</li>
+<li><code>filament</code>:                 Filament rendering engine (minimal dependencies)
+<ul>
+<li><code>backend</code>:                Rendering backends/drivers (Vulkan, Metal, OpenGL/ES)</li>
+</ul>
+</li>
+<li><code>ide</code>:                      Configuration files for IDEs (CLion, etc.)</li>
+<li><code>ios</code>:                      Sample projects for iOS</li>
+<li><code>libs</code>:                     Libraries
+<ul>
+<li><code>bluegl</code>:                 OpenGL bindings for macOS, Linux and Windows</li>
+<li><code>bluevk</code>:                 Vulkan bindings for macOS, Linux, Windows and Android</li>
+<li><code>camutils</code>:               Camera manipulation utilities</li>
+<li><code>filabridge</code>:             Library shared by the Filament engine and host tools</li>
+<li><code>filaflat</code>:               Serialization/deserialization library used for materials</li>
+<li><code>filagui</code>:                Helper library for <a href="https://github.com/ocornut/imgui">Dear ImGui</a></li>
+<li><code>filamat</code>:                Material generation library</li>
+<li><code>filamentapp</code>:            SDL2 skeleton to build sample apps</li>
+<li><code>filameshio</code>:             Tiny filamesh parsing library (see also <code>tools/filamesh</code>)</li>
+<li><code>geometry</code>:               Mesh-related utilities</li>
+<li><code>gltfio</code>:                 Loader for glTF 2.0</li>
+<li><code>ibl</code>:                    IBL generation tools</li>
+<li><code>image</code>:                  Image filtering and simple transforms</li>
+<li><code>imageio</code>:                Image file reading / writing, only intended for internal use</li>
+<li><code>matdbg</code>:                 DebugServer for inspecting shaders at run-time (debug builds only)</li>
+<li><code>math</code>:                   Math library</li>
+<li><code>mathio</code>:                 Math types support for output streams</li>
+<li><code>utils</code>:                  Utility library (threads, memory, data structures, etc.)</li>
+<li><code>viewer</code>:                 glTF viewer library (requires gltfio)</li>
+</ul>
+</li>
+<li><code>samples</code>:                  Sample desktop applications</li>
+<li><code>shaders</code>:                  Shaders used by <code>filamat</code> and <code>matc</code></li>
+<li><code>third_party</code>:              External libraries and assets
+<ul>
+<li><code>environments</code>:           Environment maps under CC0 license that can be used with <code>cmgen</code></li>
+<li><code>models</code>:                 Models under permissive licenses</li>
+<li><code>textures</code>:               Textures under CC0 license</li>
+</ul>
+</li>
+<li><code>tools</code>:                    Host tools
+<ul>
+<li><code>cmgen</code>:                  Image-based lighting asset generator</li>
+<li><code>filamesh</code>:               Mesh converter</li>
+<li><code>glslminifier</code>:           Minifies GLSL source code</li>
+<li><code>matc</code>:                   Material compiler</li>
+<li><code>filament-matp</code>:          Material parser</li>
+<li><code>matinfo</code>                 Displays information about materials compiled with <code>matc</code></li>
+<li><code>mipgen</code>                  Generates a series of miplevels from a source image</li>
+<li><code>normal-blending</code>:        Tool to blend normal maps</li>
+<li><code>resgen</code>                  Aggregates binary blobs into embeddable resources</li>
+<li><code>roughness-prefilter</code>:    Pre-filters a roughness map from a normal map to reduce aliasing</li>
+<li><code>specular-color</code>:         Computes the specular color of conductors based on spectral data</li>
+</ul>
+</li>
+<li><code>web</code>:                      JavaScript bindings, documentation, and samples</li>
+</ul>
+<h2 id="license"><a class="header" href="#license">License</a></h2>
+<p>Please see <a href="/LICENSE">LICENSE</a>.</p>
+<h2 id="disclaimer"><a class="header" href="#disclaimer">Disclaimer</a></h2>
+<p>This is not an officially supported Google product.</p>
+
+                    </main>
+
+                    <nav class="nav-wrapper" aria-label="Page navigation">
+                        <!-- Mobile navigation buttons -->
+
+                            <a rel="next prefetch" href="dup/building.html" class="mobile-nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
+                                <i class="fa fa-angle-right"></i>
+                            </a>
+
+                        <div style="clear: both"></div>
+                    </nav>
+                </div>
+            </div>
+
+            <nav class="nav-wide-wrapper" aria-label="Page navigation">
+
+                    <a rel="next prefetch" href="dup/building.html" class="nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
+                        <i class="fa fa-angle-right"></i>
+                    </a>
+            </nav>
+
+        </div>
+
+
+
+
+        <script>
+            window.playground_copyable = true;
+        </script>
+
+
+        <script src="elasticlunr.min.js"></script>
+        <script src="mark.min.js"></script>
+        <script src="searcher.js"></script>
+
+        <script src="clipboard.min.js"></script>
+        <script src="highlight.js"></script>
+        <script src="book.js"></script>
+
+        <!-- Custom JS scripts -->
+
+
+    </div>
+    </body>
 </html>

docs/main/index.html

@@ -160,8 +160,8 @@
                     <main>
                         <h1 id="core-concepts"><a class="header" href="#core-concepts">Core Concepts</a></h1>
 <ul>
-<li><a href="filament.html">Filament</a> - High-level designs; Filament's PBR/math assumptions; implementation details.</li>
-<li><a href="materials.html">Materials</a> - A guide to Filament's material definition.</li>
+<li><a href="main/filament.html">Filament</a> - High-level designs; Filament's PBR/math assumptions; implementation details.</li>
+<li><a href="main/materials.html">Materials</a> - A guide to Filament's material definition.</li>
 </ul>
 
                     </main>

docs/notes/coverage.html

@@ -250,7 +250,7 @@ execution profile data.</p>
                                 <i class="fa fa-angle-left"></i>
                             </a>
 
-                            <a rel="next prefetch" href="../notes/performance_analysis.html" class="mobile-nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
+                            <a rel="next prefetch" href="../notes/framegraph.html" class="mobile-nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
                                 <i class="fa fa-angle-right"></i>
                             </a>
 
@@ -264,7 +264,7 @@ execution profile data.</p>
                         <i class="fa fa-angle-left"></i>
                     </a>
 
-                    <a rel="next prefetch" href="../notes/performance_analysis.html" class="nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
+                    <a rel="next prefetch" href="../notes/framegraph.html" class="nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
                         <i class="fa fa-angle-right"></i>
                     </a>
             </nav>

docs/notes/framegraph.html

@@ -331,7 +331,7 @@ nodes</li>
 
                     <nav class="nav-wrapper" aria-label="Page navigation">
                         <!-- Mobile navigation buttons -->
-                            <a rel="prev" href="../notes/performance_analysis.html" class="mobile-nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
+                            <a rel="prev" href="../notes/coverage.html" class="mobile-nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
                                 <i class="fa fa-angle-left"></i>
                             </a>
 
@@ -345,7 +345,7 @@ nodes</li>
             </div>
 
             <nav class="nav-wide-wrapper" aria-label="Page navigation">
-                    <a rel="prev" href="../notes/performance_analysis.html" class="nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
+                    <a rel="prev" href="../notes/coverage.html" class="nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
                         <i class="fa fa-angle-left"></i>
                     </a>
 

docs/notes/performance_analysis.html

@@ -1,352 +0,0 @@
-<!DOCTYPE HTML>
-<html lang="en" class="light sidebar-visible" dir="ltr">
-    <head>
-        <!-- Book generated using mdBook -->
-        <meta charset="UTF-8">
-        <title>Performance analysis - Filament</title>
-
-
-        <!-- Custom HTML head -->
-
-        <meta name="description" content="">
-        <meta name="viewport" content="width=device-width, initial-scale=1">
-        <meta name="theme-color" content="#ffffff">
-
-        <link rel="shortcut icon" href="../favicon.png">
-        <link rel="stylesheet" href="../css/variables.css">
-        <link rel="stylesheet" href="../css/general.css">
-        <link rel="stylesheet" href="../css/chrome.css">
-
-        <!-- Fonts -->
-        <link rel="stylesheet" href="../FontAwesome/css/font-awesome.css">
-        <link rel="stylesheet" href="../fonts/fonts.css">
-
-        <!-- Highlight.js Stylesheets -->
-        <link rel="stylesheet" href="../highlight.css">
-        <link rel="stylesheet" href="../tomorrow-night.css">
-        <link rel="stylesheet" href="../ayu-highlight.css">
-
-        <!-- Custom theme stylesheets -->
-
-        <!-- MathJax -->
-        <script async src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
-
-        <!-- Provide site root to javascript -->
-        <script>
-            var path_to_root = "../";
-            var default_theme = window.matchMedia("(prefers-color-scheme: dark)").matches ? "light" : "light";
-        </script>
-        <!-- Start loading toc.js asap -->
-        <script src="../toc.js"></script>
-    </head>
-    <body>
-    <div id="body-container">
-        <!-- Work around some values being stored in localStorage wrapped in quotes -->
-        <script>
-            try {
-                var theme = localStorage.getItem('mdbook-theme');
-                var sidebar = localStorage.getItem('mdbook-sidebar');
-
-                if (theme.startsWith('"') && theme.endsWith('"')) {
-                    localStorage.setItem('mdbook-theme', theme.slice(1, theme.length - 1));
-                }
-
-                if (sidebar.startsWith('"') && sidebar.endsWith('"')) {
-                    localStorage.setItem('mdbook-sidebar', sidebar.slice(1, sidebar.length - 1));
-                }
-            } catch (e) { }
-        </script>
-
-        <!-- Set the theme before any content is loaded, prevents flash -->
-        <script>
-            var theme;
-            try { theme = localStorage.getItem('mdbook-theme'); } catch(e) { }
-            if (theme === null || theme === undefined) { theme = default_theme; }
-            const html = document.documentElement;
-            html.classList.remove('light')
-            html.classList.add(theme);
-            html.classList.add("js");
-        </script>
-
-        <input type="checkbox" id="sidebar-toggle-anchor" class="hidden">
-
-        <!-- Hide / unhide sidebar before it is displayed -->
-        <script>
-            var sidebar = null;
-            var sidebar_toggle = document.getElementById("sidebar-toggle-anchor");
-            if (document.body.clientWidth >= 1080) {
-                try { sidebar = localStorage.getItem('mdbook-sidebar'); } catch(e) { }
-                sidebar = sidebar || 'visible';
-            } else {
-                sidebar = 'hidden';
-            }
-            sidebar_toggle.checked = sidebar === 'visible';
-            html.classList.remove('sidebar-visible');
-            html.classList.add("sidebar-" + sidebar);
-        </script>
-
-        <nav id="sidebar" class="sidebar" aria-label="Table of contents">
-            <div style="display:flex;align-items:center;justify-content:center">
-                <img class="flogo" src="../images/filament_logo_small.png"></img>
-            </div>
-            <!-- populated by js -->
-            <mdbook-sidebar-scrollbox class="sidebar-scrollbox"></mdbook-sidebar-scrollbox>
-            <noscript>
-                <iframe class="sidebar-iframe-outer" src="../toc.html"></iframe>
-            </noscript>
-            <div id="sidebar-resize-handle" class="sidebar-resize-handle">
-                <div class="sidebar-resize-indicator"></div>
-            </div>
-        </nav>
-
-        <div id="page-wrapper" class="page-wrapper">
-
-            <div class="page">
-                <div id="menu-bar-hover-placeholder"></div>
-                <div id="menu-bar" class="menu-bar sticky">
-                    <div class="left-buttons">
-                        <label id="sidebar-toggle" class="icon-button" for="sidebar-toggle-anchor" title="Toggle Table of Contents" aria-label="Toggle Table of Contents" aria-controls="sidebar">
-                            <i class="fa fa-bars"></i>
-                        </label>
-                        <!-- Filament: disable themes because the markdeep part does not look good for dark themes -->
-                        <!--
-                        <button id="theme-toggle" class="icon-button" type="button" title="Change theme" aria-label="Change theme" aria-haspopup="true" aria-expanded="false" aria-controls="theme-list">
-                            <i class="fa fa-paint-brush"></i>
-                        </button>
-                        <ul id="theme-list" class="theme-popup" aria-label="Themes" role="menu">
-                            <li role="none"><button role="menuitem" class="theme" id="light">Light</button></li>
-                            <li role="none"><button role="menuitem" class="theme" id="rust">Rust</button></li>
-                            <li role="none"><button role="menuitem" class="theme" id="coal">Coal</button></li>
-                            <li role="none"><button role="menuitem" class="theme" id="navy">Navy</button></li>
-                            <li role="none"><button role="menuitem" class="theme" id="ayu">Ayu</button></li>
-                        </ul>
-                        -->
-                        <button id="search-toggle" class="icon-button" type="button" title="Search. (Shortkey: s)" aria-label="Toggle Searchbar" aria-expanded="false" aria-keyshortcuts="S" aria-controls="searchbar">
-                            <i class="fa fa-search"></i>
-                        </button>
-                    </div>
-
-                    <h1 class="menu-title">Filament</h1>
-
-                    <div class="right-buttons">
-                        <a href="https://github.com/google/filament" title="Git repository" aria-label="Git repository">
-                            <i id="git-repository-button" class="fa fa-github"></i>
-                        </a>
-
-                    </div>
-                </div>
-
-                <div id="search-wrapper" class="hidden">
-                    <form id="searchbar-outer" class="searchbar-outer">
-                        <input type="search" id="searchbar" name="searchbar" placeholder="Search this book ..." aria-controls="searchresults-outer" aria-describedby="searchresults-header">
-                    </form>
-                    <div id="searchresults-outer" class="searchresults-outer hidden">
-                        <div id="searchresults-header" class="searchresults-header"></div>
-                        <ul id="searchresults">
-                        </ul>
-                    </div>
-                </div>
-
-                <!-- Apply ARIA attributes after the sidebar and the sidebar toggle button are added to the DOM -->
-                <script>
-                    document.getElementById('sidebar-toggle').setAttribute('aria-expanded', sidebar === 'visible');
-                    document.getElementById('sidebar').setAttribute('aria-hidden', sidebar !== 'visible');
-                    Array.from(document.querySelectorAll('#sidebar a')).forEach(function(link) {
-                        link.setAttribute('tabIndex', sidebar === 'visible' ? 0 : -1);
-                    });
-                </script>
-
-                <div id="content" class="content">
-                    <main>
-                        <h1 id="performance-analysis"><a class="header" href="#performance-analysis">Performance Analysis</a></h1>
-<h2 id="android"><a class="header" href="#android">Android</a></h2>
-<h3 id="prerequisites"><a class="header" href="#prerequisites">Prerequisites</a></h3>
-<ul>
-<li>Download and install Android GPU Inspector (AGI). See https://developer.android.com/agi.</li>
-</ul>
-<hr />
-<h4 id="profiling"><a class="header" href="#profiling">Profiling</a></h4>
-<ol>
-<li>
-<p>Before profiling the application or analyzing the performance in a consistent way, ideally
-the <strong>GPU frequency on the target hardware should get locked</strong>. In order to do this, you need to do the following:</p>
-<ol>
-<li>Ensure your device is OEM unlocked. If your phone is carrier-locked, you may need
-to wait a period, such as 60 or 90 days after activation to be eligible for unlocking.
-Some phones don't support this at all. You will need to enable developer options
-(e.g. Settings &gt; About Phone and tap the Build number 7 times). You need to go to
-Settings &gt; System &gt; Developer options and toggle on "OEM unlocking".</li>
-<li>Next, you need to unlock the phone. You will need to install Android SDK Platform
-Tools on you computer, enable "USB debugging" in your phone's Settings &gt; System &gt; Developer options.</li>
-<li>Connect the phone to the computer via a USB cable and run from the command line:
-<pre><code class="language-shell">adb reboot bootloader
-# once the phone is in bootloader mode, run the following to begin the unlocking
-# process:
-fastboot flashing unlock
-</code></pre>
-A warning will appear on your phone's screen. Use the volume buttons to navigate and
-the power button to select the "Unlock the bootloader" option. The phone will perform
-a factory data reset and reboot with an unlocked bootloader.</li>
-<li>You would need to flash an image to the phone with root permissions, such as a *-userdebug or *-eng
-build. One way to do this is with the <a href="https://flash.android.com/">Android Flash Tool</a>. Connect the
-tool to your device and find a build to flash ending in <code>-userdebug</code> or <code>-eng</code>. Once you have that
-selected run <code>Install build</code>.</li>
-<li>Shell into your device as root and configure the gpu frequency to be locked, e.g.:
-<pre><code class="language-shell">adb shell
-su
-# navigate to the system GPU directory. this varies on different phones. One phone might have
-# it at /sys/class/kgsl/kgsl-3d0 and another might be in something similar, maybe with "mali" instead
-# of "kgsl". At the time of writing this for the device at hand it was /sys/devices/platform/1f000000.mali
-cd /sys/devices/platform/1f000000.mali
-# get the current available GPU governors and frequencies.
-# note that some systems may have these at gpu_available_governors and gpu_available_frequencies, but
-# the system at the time of writing this had available_governors and available_frequences
-cat available_governors
-cat available_frequencies
-# depending on the governors, you may want to set it prioritize performance over other things like
-# battery. Some systems allow you to do this with something like (although, for the device used at the
-# time of writing this did not have an equivalent option):
-echo performance &gt; gpu_governor
-# finally, lock your frequency in, usually to something high like 897 MHz. Some systems may have you
-# pipe the value to gpu_min_freq and gpu_max_freq, but the system used at the time of writing this had:
-echo 940000 &gt; hint_min_freq
-echo 940000 &gt; hint_max_freq
-# you can typically verify the GPU is running at that frequency consistently by running something like
-# the following a few times over time, which should show the frequency you want to lock the device to:
-cat cur_freq
-</code></pre>
-</li>
-<li>You may need to re-apply the <code>hint_min_freq</code> just before starting the profiling trace and check before and after
-that the frequency remained at the value expected. Some systems may adjust the frequency on you, but you
-may want to ensure the frequency remains the same through the analysis.</li>
-<li>The GPU frequency settings should be undone after restarting the device, but after you have done your app profiling,
-you can revert the state of the device, such as the OS build image, back to the way you had it
-initially, as needed.</li>
-</ol>
-</li>
-<li>
-<p>Build a release build of Filament with the applicable backend(s) enabled
-<em>(+ any special flags for enabling sys strace. Nothing special is needed for Vulkan or WebGPU aside from
-building a release build with no flags)</em> <em>(debug builds for this are useless)</em></p>
-<pre><code class="language-shell"># the following command assumes you are in the root filament directory
-# and ANDROID_HOME is exported (and possibly also CC and CXX on linux as needed)
-#
-# NOTE: to build with WebGPU support you need to explicitly include the -W flag
-# (it doesn't get compiled in by default), e.g.:
-# ./build.sh -W -p android,desktop -i release
-#
-# Note that you can speed this up a bit (and reduce disk space usage) by limiting the target to just 
-# the ABI you plan on testing with the -q flag, e.g. -q arm64-v8a. If you do this, you
-# will need to update the android/gradle.properties file to specify the ABI(s) you are targeting
-# with the com.google.android.filament.abis property.
-# Thus, a build command that would target BOTH Vulkan AND WebGPU AND only target the ARM64 ABI would look something
-# like:
-# ./build.sh -W -q arm64-v8a -p android,desktop -i release
-./build.sh -p android,desktop -i release
-</code></pre>
-</li>
-<li>
-<p>Connect your Android device to your computer via a USB cord with USB debugging enabled and configure
-the system property to default to the desired backend, e.g.
-<em>(to determine how these numbers map to the backends, see the <code>enum class Backend</code>
-definition in <a href="../../../../filament/backend/include/backend/DriverEnums.h">filament/filament/backend/include/backend/DriverEnums.h</a>)</em>:</p>
-<pre><code class="language-shell"># to set the backend to Vulkan: 
-adb shell setprop debug.filament.backend 2
-# to set the backend to WebGPU:
-adb shell setprop debug.filament.backend 4
-# to view the current property:
-adb shell getprop debug.filament.backend
-</code></pre>
-</li>
-<li>
-<p>Build and run a sample, e.g. sample-gltf-viewer, on your Android device
-<em>(<a href="https://developer.android.com/studio">Android Studio</a> recommended)</em>.</p>
-</li>
-<li>
-<p>Run <a href="https://developer.android.com/agi">AGI</a> and follow instructions to profile the app/system
-with trace capture(s). See https://developer.android.com/agi/start for more details to
-get started.</p>
-<ul>
-<li>We typically only run "Capture System Profiler trace" <em>(not necessarily "Capture Frame Profiler trace")</em></li>
-<li>When configuring the trace:
-<ul>
-<li>For both the WebGPU and Vulkan backends configure the profiler for the Vulkan API
-<em>(since WebGPU should be using Vulkan under the hood as well)</em></li>
-<li>Running for ~1 seconds should suffice</li>
-<li>Hit the "Configure" button in Trace objects, select "Switch to advanced mode" and add:
-<pre><code>data_sources {
-  config {
-    name: "track_event"
-    track_event_config {
-      disabled_categories: "*"
-      enabled_categories: "filament/filament"
-      enabled_categories: "filament/jobsystem"
-      enabled_categories: "filament/gltfio"
-    }
-  }
-}
-</code></pre>
-</li>
-</ul>
-</li>
-<li>One you open the trace, zoom into a series of frames to get a sense of generally how long they typically take
-<em>(use <code>W</code>, <code>S</code>, <code>A</code> and <code>D</code> keys and mouse wheel for navigation)</em> and find a representative one.
-We are most interested in the performance of the
-<code>FEngine::loop</code> thread, how long it takes, overlap in activities/processes/commands, reduction in queue submissions,
-etc. Similarly, we can view GPU timeline as it relates to that. We want to see overlapping
-shader invocations and non-interrupted fragment shader runs.</li>
-</ul>
-</li>
-</ol>
-
-                    </main>
-
-                    <nav class="nav-wrapper" aria-label="Page navigation">
-                        <!-- Mobile navigation buttons -->
-                            <a rel="prev" href="../notes/coverage.html" class="mobile-nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
-                                <i class="fa fa-angle-left"></i>
-                            </a>
-
-                            <a rel="next prefetch" href="../notes/framegraph.html" class="mobile-nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
-                                <i class="fa fa-angle-right"></i>
-                            </a>
-
-                        <div style="clear: both"></div>
-                    </nav>
-                </div>
-            </div>
-
-            <nav class="nav-wide-wrapper" aria-label="Page navigation">
-                    <a rel="prev" href="../notes/coverage.html" class="nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
-                        <i class="fa fa-angle-left"></i>
-                    </a>
-
-                    <a rel="next prefetch" href="../notes/framegraph.html" class="nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
-                        <i class="fa fa-angle-right"></i>
-                    </a>
-            </nav>
-
-        </div>
-
-
-
-
-        <script>
-            window.playground_copyable = true;
-        </script>
-
-
-        <script src="../elasticlunr.min.js"></script>
-        <script src="../mark.min.js"></script>
-        <script src="../searcher.js"></script>
-
-        <script src="../clipboard.min.js"></script>
-        <script src="../highlight.js"></script>
-        <script src="../book.js"></script>
-
-        <!-- Custom JS scripts -->
-
-
-    </div>
-    </body>
-</html>

docs_src/README.md

@@ -65,9 +65,6 @@ We describe step 1 in detail for the sake of record:
  - Replace css styling in the exported output as needed (so they don't interfere with the book's css.
  - Replace resource urls to refer to locations relative to the mdbook structure.
 
-Any `markdeep` doc can be placed in `docs_src/src_markdeep/` and they will be parsed to html and included
-in the book as above.
-
 ### READMEs
 Filament depends on a number of libraries, which reside in the directory `libs`. These individual
 libaries often have README.md in their root to describe itself. We collect these descriptions into our
@@ -78,7 +75,7 @@ The process for copying and processing these READMEs is outlined in [Introductor
 
 ### Other technical notes
 These are technical documents that do not fit into a library, tool, or directory of the
-Filament source tree. We collect them into the `docs_src/src_mdbook/src/notes` directory. No additional
+Filament source tree. We collect them into the `docs_src/src/notes` directory. No additional
 processing is needed for these documents.
 
 ### Raw source files
@@ -93,7 +90,7 @@ add a link in `SUMMARY.md`, and perform the steps outlined in
 [how-to create section](#how-to-create).
 
 For example, if you are adding a general technical note, then you would
- - Place the document (file with extension `.md`) in `docs_src/src_mdbook/src/notes`
+ - Place the document (file with extension `.md`) in `docs_src/src/notes`
  - Add a link in [`docs_src/src_mdbook/src/SUMMARY.md`]
  - Run the commands in the [Generate](#how-to-generate) section
 

docs_src/build/duplicates.json

@@ -2,14 +2,9 @@
     "README.md": {
         "dest": "dup/intro.md",
         "link_transforms": {
-            "/BUILDING.md#running-the-native-samples": "building.md#running-the-native-samples",
-            "/BUILDING.md": "building.md",
+            "BUILDING.md": "building.md",
             "/CONTRIBUTING.md": "contributing.md",
             "/CODE_STYLE.md": "code_style.md",
-            "/LICENSE": "https://github.com/google/filament/blob/main/LICENSE",
-            "https://google.github.io/filament/Filament.html": "../main/filament.md",
-            "https://google.github.io/filament/Materials.html": "../main/materials.md",
-            "https://google.github.io/filament/notes/material_properties.html": "../notes/material_properties.md",
             "docs/images/samples": "../images/samples"
         }
     },

docs_src/src_mdbook/src/SUMMARY.md

@@ -24,7 +24,6 @@
     - [Running with ASAN and UBSAN](./notes/asan_ubsan.md)
     - [Using Instruments on macOS](./notes/instruments.md)
     - [Code coverage analysis](./notes/coverage.md)
-    - [Performance analysis](./notes/performance_analysis.md)
   - [Framegraph](./notes/framegraph.md)
   - [Libraries](./notes/libs.md)
     - [bluegl](./dup/bluegl.md)

docs_src/src_mdbook/src/main/README.md

@@ -1,5 +1,5 @@
 # Core Concepts
 
 
-- [Filament](filament.md) - High-level designs; Filament's PBR/math assumptions; implementation details.
-- [Materials](materials.md) - A guide to Filament's material definition.
+- [Filament](main/filament.md) - High-level designs; Filament's PBR/math assumptions; implementation details.
+- [Materials](main/materials.md) - A guide to Filament's material definition.

docs_src/src_mdbook/src/notes/performance_analysis.md

@@ -1,129 +0,0 @@
-# Performance Analysis
-
-## Android
-
-### Prerequisites
-
-- Download and install Android GPU Inspector (AGI). See https://developer.android.com/agi.
-
----
-
-#### Profiling
-
-1. Before profiling the application or analyzing the performance in a consistent way, ideally
-   the **GPU frequency on the target hardware should get locked**. In order to do this, you need to do the following:
-   1. Ensure your device is OEM unlocked. If your phone is carrier-locked, you may need
-      to wait a period, such as 60 or 90 days after activation to be eligible for unlocking.
-      Some phones don't support this at all. You will need to enable developer options
-      (e.g. Settings > About Phone and tap the Build number 7 times). You need to go to
-      Settings > System > Developer options and toggle on "OEM unlocking".
-   1. Next, you need to unlock the phone. You will need to install Android SDK Platform
-      Tools on you computer, enable "USB debugging" in your phone's Settings > System > Developer options.
-   1. Connect the phone to the computer via a USB cable and run from the command line:
-      ```shell
-      adb reboot bootloader
-      # once the phone is in bootloader mode, run the following to begin the unlocking
-      # process:
-      fastboot flashing unlock
-      ```
-      A warning will appear on your phone's screen. Use the volume buttons to navigate and
-      the power button to select the "Unlock the bootloader" option. The phone will perform
-      a factory data reset and reboot with an unlocked bootloader.
-   1. You would need to flash an image to the phone with root permissions, such as a *-userdebug or *-eng
-      build. One way to do this is with the [Android Flash Tool](https://flash.android.com/). Connect the
-      tool to your device and find a build to flash ending in `-userdebug` or `-eng`. Once you have that
-      selected run `Install build`.
-   1. Shell into your device as root and configure the gpu frequency to be locked, e.g.:
-      ```shell
-      adb shell
-      su
-      # navigate to the system GPU directory. this varies on different phones. One phone might have
-      # it at /sys/class/kgsl/kgsl-3d0 and another might be in something similar, maybe with "mali" instead
-      # of "kgsl". At the time of writing this for the device at hand it was /sys/devices/platform/1f000000.mali
-      cd /sys/devices/platform/1f000000.mali
-      # get the current available GPU governors and frequencies.
-      # note that some systems may have these at gpu_available_governors and gpu_available_frequencies, but
-      # the system at the time of writing this had available_governors and available_frequences
-      cat available_governors
-      cat available_frequencies
-      # depending on the governors, you may want to set it prioritize performance over other things like
-      # battery. Some systems allow you to do this with something like (although, for the device used at the
-      # time of writing this did not have an equivalent option):
-      echo performance > gpu_governor
-      # finally, lock your frequency in, usually to something high like 897 MHz. Some systems may have you
-      # pipe the value to gpu_min_freq and gpu_max_freq, but the system used at the time of writing this had:
-      echo 940000 > hint_min_freq
-      echo 940000 > hint_max_freq
-      # you can typically verify the GPU is running at that frequency consistently by running something like
-      # the following a few times over time, which should show the frequency you want to lock the device to:
-      cat cur_freq
-      ```
-   1. You may need to re-apply the `hint_min_freq` just before starting the profiling trace and check before and after
-      that the frequency remained at the value expected. Some systems may adjust the frequency on you, but you
-      may want to ensure the frequency remains the same through the analysis.
-   1. The GPU frequency settings should be undone after restarting the device, but after you have done your app profiling,
-      you can revert the state of the device, such as the OS build image, back to the way you had it
-      initially, as needed.
-1. Build a release build of Filament with the applicable backend(s) enabled
-   _(+ any special flags for enabling sys strace. Nothing special is needed for Vulkan or WebGPU aside from
-   building a release build with no flags)_ _(debug builds for this are useless)_
-
-    ```shell
-    # the following command assumes you are in the root filament directory
-    # and ANDROID_HOME is exported (and possibly also CC and CXX on linux as needed)
-    #
-    # NOTE: to build with WebGPU support you need to explicitly include the -W flag
-    # (it doesn't get compiled in by default), e.g.:
-    # ./build.sh -W -p android,desktop -i release
-    #
-    # Note that you can speed this up a bit (and reduce disk space usage) by limiting the target to just 
-    # the ABI you plan on testing with the -q flag, e.g. -q arm64-v8a. If you do this, you
-    # will need to update the android/gradle.properties file to specify the ABI(s) you are targeting
-    # with the com.google.android.filament.abis property.
-    # Thus, a build command that would target BOTH Vulkan AND WebGPU AND only target the ARM64 ABI would look something
-    # like:
-    # ./build.sh -W -q arm64-v8a -p android,desktop -i release
-    ./build.sh -p android,desktop -i release
-    ```
-1. Connect your Android device to your computer via a USB cord with USB debugging enabled and configure
-   the system property to default to the desired backend, e.g.
-   _(to determine how these numbers map to the backends, see the `enum class Backend`
-   definition in [filament/filament/backend/include/backend/DriverEnums.h](../../../../filament/backend/include/backend/DriverEnums.h))_:
-    ```shell
-    # to set the backend to Vulkan: 
-    adb shell setprop debug.filament.backend 2
-    # to set the backend to WebGPU:
-    adb shell setprop debug.filament.backend 4
-    # to view the current property:
-    adb shell getprop debug.filament.backend
-    ```
-1. Build and run a sample, e.g. sample-gltf-viewer, on your Android device
-   _([Android Studio](https://developer.android.com/studio) recommended)_.
-1. Run [AGI](https://developer.android.com/agi) and follow instructions to profile the app/system
-   with trace capture(s). See https://developer.android.com/agi/start for more details to
-   get started.
-   - We typically only run "Capture System Profiler trace" _(not necessarily "Capture Frame Profiler trace")_
-   - When configuring the trace:
-     - For both the WebGPU and Vulkan backends configure the profiler for the Vulkan API
-       _(since WebGPU should be using Vulkan under the hood as well)_
-     - Running for ~1 seconds should suffice
-     - Hit the "Configure" button in Trace objects, select "Switch to advanced mode" and add:
-       ```
-       data_sources {
-         config {
-           name: "track_event"
-           track_event_config {
-             disabled_categories: "*"
-             enabled_categories: "filament/filament"
-             enabled_categories: "filament/jobsystem"
-             enabled_categories: "filament/gltfio"
-           }
-         }
-       }
-       ```
-   - One you open the trace, zoom into a series of frames to get a sense of generally how long they typically take
-    _(use `W`, `S`, `A` and `D` keys and mouse wheel for navigation)_ and find a representative one.
-     We are most interested in the performance of the
-     `FEngine::loop` thread, how long it takes, overlap in activities/processes/commands, reduction in queue submissions,
-     etc. Similarly, we can view GPU timeline as it relates to that. We want to see overlapping
-     shader invocations and non-interrupted fragment shader runs.

docs_src/src_raw/index.html

@@ -1,11 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-<head>
-    <meta charset="UTF-8">
-    <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Redirecting...</title>
-    <meta http-equiv="refresh" content="0; url=./dup/intro.html">
-</head>
-<body>
-</body>
-</html>

filament/backend/CMakeLists.txt

@@ -258,8 +258,6 @@ if (FILAMENT_SUPPORTS_WEBGPU)
             include/backend/platforms/WebGPUPlatform.h
             src/webgpu/platform/WebGPUPlatform.cpp
             src/webgpu/SpdMipmapGenerator/SpdMipmapGenerator.cpp
-            src/webgpu/utils/AsyncTaskCounter.cpp
-            src/webgpu/utils/AsyncTaskCounter.h
             src/webgpu/utils/StringPlaceholderTemplateProcessor.cpp
             src/webgpu/utils/StringPlaceholderTemplateProcessor.h
             src/webgpu/WebGPUBlitter.cpp
@@ -414,7 +412,6 @@ endif()
 
 if (APPLE AND NOT IOS)
     target_link_libraries(${TARGET} PRIVATE "-framework Cocoa")
-    target_link_libraries(${TARGET} PRIVATE "-framework QuartzCore")
 endif()
 
 target_link_libraries(${TARGET} PUBLIC math)
@@ -530,7 +527,7 @@ target_link_libraries(${TARGET} PRIVATE
 if (LINUX AND FILAMENT_SUPPORTS_WEBGPU)
     # Unclear why this needs to be explicit while other linux libs do not, but it is necessary for
     # WebGPU to build on linux.
-    target_link_libraries(${TARGET} PRIVATE -lxcb -lX11)
+    target_link_libraries(${TARGET} PRIVATE -lxcb)
 endif()
 
 # ==================================================================================================
@@ -579,11 +576,6 @@ if (APPLE OR LINUX)
         test/test_CommandBufferQueue.cpp
         test/test_Template.cpp
     )
-    if (FILAMENT_SUPPORTS_WEBGPU)
-        list(APPEND BACKEND_TEST_SRC
-                test/test_WebGPUAsyncTaskCounter.cpp
-                test/test_WebGPUComposeSwizzle.cpp)
-    endif()
     if (APPLE)
         # Metal-specific tests
         list(APPEND BACKEND_TEST_SRC
@@ -601,7 +593,6 @@ if (APPLE OR LINUX)
         spirv-cross-glsl)
     # Create input/output directories for test result images.
     file(MAKE_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/images/actual_images)
-    file(MAKE_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/images/diff_images)
     file(COPY test/expected_images DESTINATION ${CMAKE_CURRENT_BINARY_DIR}/images)
 endif()
 
@@ -631,10 +622,6 @@ if (APPLE AND NOT IOS)
         target_link_libraries(backend_test PRIVATE image imageio)
         list(APPEND BACKEND_TEST_DEPS image)
     endif()
-    if (FILAMENT_SUPPORTS_WEBGPU)
-        target_link_libraries(backend_test PRIVATE webgpu_dawn dawncpp_headers)
-        list(APPEND BACKEND_TEST_DEPS webgpu_dawn dawncpp_headers)
-    endif()
 
     set(BACKEND_TEST_COMBINED_OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/libbackendtest_combined.a")
     combine_static_libs(backend_test "${BACKEND_TEST_COMBINED_OUTPUT}" "${BACKEND_TEST_DEPS}")
@@ -666,9 +653,6 @@ if (LINUX)
     add_executable(backend_test_linux test/linux_runner.cpp ${BACKEND_TEST_SRC})
     target_link_libraries(backend_test_linux PRIVATE ${BACKEND_TEST_LIBS})
     set_target_properties(backend_test_linux PROPERTIES FOLDER Tests)
-    if (FILAMENT_SUPPORTS_WEBGPU)
-        target_link_libraries(backend_test_linux PRIVATE webgpu_dawn dawncpp_headers)
-    endif()
 endif()
 
 # ==================================================================================================

filament/backend/include/backend/DriverEnums.h

@@ -629,18 +629,6 @@ enum class PrimitiveType : uint8_t {
     TRIANGLE_STRIP = 5     //!< triangle strip
 };
 
-[[nodiscard]] constexpr bool isStripPrimitiveType(const PrimitiveType type) {
-    switch (type) {
-        case PrimitiveType::POINTS:
-        case PrimitiveType::LINES:
-        case PrimitiveType::TRIANGLES:
-            return false;
-        case PrimitiveType::LINE_STRIP:
-        case PrimitiveType::TRIANGLE_STRIP:
-            return true;
-    }
-}
-
 /**
  * Supported uniform types
  */

filament/backend/include/backend/Platform.h

@@ -156,7 +156,7 @@ public:
          * Set to `true` to forcibly disable amortized shader compilation in the backend.
          * Currently only honored by the GL backend.
          */
-        bool disableAmortizedShaderCompile = true;
+        bool disableAmortizedShaderCompile = false;
 
         /**
          * Disable backend handles use-after-free checks.

filament/backend/include/backend/platforms/WebGPUPlatform.h

@@ -19,16 +19,6 @@
 
 #include <backend/Platform.h>
 
-#if defined(__linux__) && defined(FILAMENT_SUPPORTS_X11)
-// Resolve the conflicts between webgpu_cpp.h and X11 defines
-#undef Always
-#undef Success
-#undef None
-#undef True
-#undef False
-#undef Status
-#undef Bool
-#endif
 #include <webgpu/webgpu_cpp.h>
 
 #include <cstdint>
@@ -40,7 +30,7 @@ namespace filament::backend {
  * A Platform interface, handling the environment-specific concerns, e.g. OS, for creating a WebGPU
  * driver (backend).
  */
-class WebGPUPlatform : public Platform {
+class WebGPUPlatform final : public Platform {
 public:
     WebGPUPlatform();
     ~WebGPUPlatform() override = default;
@@ -62,14 +52,6 @@ public:
     // either returns a valid device or panics
     [[nodiscard]] wgpu::Device requestDevice(wgpu::Adapter const& adapter);
 
-    struct Configuration {
-        wgpu::BackendType forceBackendType =  wgpu::BackendType::Undefined;
-    };
-
-    [[nodiscard]] virtual Configuration getConfiguration() const noexcept {
-        return {};
-    }
-
 protected:
     [[nodiscard]] Driver* createDriver(void* sharedContext,
             const Platform::DriverConfig& driverConfig) noexcept override;

filament/backend/src/CommandStream.cpp

@@ -97,13 +97,11 @@ void CommandStream::execute(void* buffer) {
         }
     }
 
-    Driver& UTILS_RESTRICT driver = mDriver;
-    CommandBase* UTILS_RESTRICT base = static_cast<CommandBase*>(buffer);
-    mDriver.execute([&driver, base] {
-        auto& d = driver;
-        auto p = base;
-        while (UTILS_LIKELY(p)) {
-            p = p->execute(d);
+    mDriver.execute([this, buffer]() {
+        Driver& UTILS_RESTRICT driver = mDriver;
+        CommandBase* UTILS_RESTRICT base = static_cast<CommandBase*>(buffer);
+        while (UTILS_LIKELY(base)) {
+            base = base->execute(driver);
         }
     });
 

filament/backend/src/metal/MetalHandles.mm

@@ -118,10 +118,6 @@ MetalSwapChain::MetalSwapChain(
       layerDrawableMutex(std::make_shared<std::mutex>()),
       type(SwapChainType::CAMETALLAYER) {
 
-    FILAMENT_CHECK_PRECONDITION([nativeWindow isKindOfClass:[CAMetalLayer class]])
-            << "nativeWindow pointer of class "
-            << [NSStringFromClass([nativeWindow class]) UTF8String] << " is not a CAMetalLayer";
-
     if (!(flags & SwapChain::CONFIG_TRANSPARENT) && !nativeWindow.opaque) {
         LOG(WARNING) << "Warning: Filament SwapChain has no CONFIG_TRANSPARENT flag, but the "
                         "CAMetaLayer("

filament/backend/src/opengl/OpenGLDriver.cpp

@@ -1948,7 +1948,7 @@ void OpenGLDriver::destroyTexture(Handle<HwTexture> th) {
                     if (UTILS_UNLIKELY(t->hwStream)) {
                         detachStream(t);
                     }
-                    if (UTILS_UNLIKELY(t->externalTexture)) {
+                    if (UTILS_UNLIKELY(t->target == SamplerType::SAMPLER_EXTERNAL)) {
                         mPlatform.destroyExternalImageTexture(t->externalTexture);
                     } else {
                         glDeleteTextures(1, &t->gl.id);

filament/backend/src/vulkan/VulkanCommands.cpp

@@ -211,9 +211,7 @@ VkSemaphore VulkanCommandBuffer::submit() {
     FVK_LOGI << "Submitting cmdbuffer=" << mBuffer
              << " wait=(";
     for (size_t s = 0, count = mWaitSemaphores.size(); s < count; ++s) {
-        FVK_LOGI << "\tsemaphore=" << mWaitSemaphores[s] << "|stage="
-                 << mWaitSemaphoreStages[s]
-                 << (s <  mWaitSemaphores.size() - 1 ? "\n" : "");
+        FVK_LOGI << mWaitSemaphores[s] << " ";
     }
     FVK_LOGI << ") "
              << " signal=" << mSubmission
@@ -436,13 +434,10 @@ bool VulkanCommands::flush() {
         }
         if (lastSubmit != VK_NULL_HANDLE) {
             // Note that the stage we're waiting on is the fragment shader stage.  This assumes
-            // that the subsequent command buffer will only depend on
-            //    1) fragment output of the previous command buffer
-            //    2) reading/writing of buffers (i.e. UBO) of the previous command buffer
-            // Restricting the wait stages will allow for vertex work to proceed (more overlapping
+            // that the subsequent command buffer will only dependent on the fragment output of the
+            // previous buffer, allowing for vertex work to proceed (more overlapping
             // vertex/fragment work).
-            pool->waitFor(lastSubmit,
-                    VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT | VK_PIPELINE_STAGE_TRANSFER_BIT);
+            pool->waitFor(lastSubmit, VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT);
             lastSubmit = VK_NULL_HANDLE;
         }
         dependency = pool->flush();

filament/backend/src/vulkan/VulkanDriver.cpp

@@ -452,7 +452,7 @@ void VulkanDriver::updateDescriptorSetTexture(
     auto set = resource_ptr<VulkanDescriptorSet>::cast(&mResourceManager, dsh);
     auto texture = resource_ptr<VulkanTexture>::cast(&mResourceManager, th);
 
-    if (UTILS_UNLIKELY(mExternalImageManager.isExternallySampledTexture(texture))) {
+    if (mExternalImageManager.isExternallySampledTexture(texture)) {
         mExternalImageManager.bindExternallySampledTexture(set, binding, texture, params);
         mAppState.hasBoundExternalImages = true;
     } else {
@@ -1939,18 +1939,6 @@ void VulkanDriver::bindDescriptorSet(
     if (dsh) {
         auto set = resource_ptr<VulkanDescriptorSet>::cast(&mResourceManager, dsh);
         mDescriptorSetCache.bind(setIndex, set, std::move(offsets));
-
-        if (mAppState.hasExternalSamplers()) {
-            auto const& bindInDrawBundle = mPipelineState.bindInDraw.second;
-            // The set index being bound has already been bound or will be bound. If it's already
-            // been bound and this set has external samplers, we do the doBindindraw block in
-            // draw2() again. Because this set might potentially cause a new pipelineLayout
-            // (therefore pipeline) to be bound.
-            if (bindInDrawBundle.descriptorSetMask[setIndex] &&
-                    mExternalImageManager.hasExternalSampler(set)) {
-                mPipelineState.bindInDraw.first = true;
-            }
-        }
     } else {
         mDescriptorSetCache.unbind(setIndex);
     }

filament/backend/src/vulkan/VulkanExternalImageManager.cpp

@@ -121,7 +121,7 @@ fvkutils::DescriptorSetMask VulkanExternalImageManager::prepareBindSets(LayoutAr
 }
 
 bool VulkanExternalImageManager::hasExternalSampler(
-        fvkmemory::resource_ptr<VulkanDescriptorSet> set) const {
+        fvkmemory::resource_ptr<VulkanDescriptorSet> set) {
     auto itr = std::find_if(mSetBindings.begin(), mSetBindings.end(),
             [&](SetBindingInfo const& info) { return info.set == set; });
     return itr != mSetBindings.end();
@@ -191,11 +191,9 @@ void VulkanExternalImageManager::updateSetAndLayout(
         VkDescriptorSet const srcSet = oldSet != VK_NULL_HANDLE ? oldSet : set->getVkSet();
         copySet(mPlatform->getDevice(), srcSet, newSet, copyBindings);
 
-        set->setExternalSamplerVkSet(newSet,
-                [&descriptorSetCache = mDescriptorSetCache, layoutCount = layout->count, newLayout,
-                        newSet](VulkanDescriptorSet*) {
-                    descriptorSetCache->manualRecycle(layoutCount, newLayout, newSet);
-                });
+        set->setExternalSamplerVkSet(newSet, [&](VulkanDescriptorSet*) {
+            mDescriptorSetCache->manualRecycle(layout->count, newLayout, newSet);
+        });
         if (oldLayout != newLayout) {
             layout->setExternalSamplerVkLayout(newLayout);
         }

filament/backend/src/vulkan/VulkanExternalImageManager.h

@@ -86,9 +86,9 @@ public:
         VkSamplerYcbcrConversion conversion = VK_NULL_HANDLE;
     };
 
-    bool hasExternalSampler(fvkmemory::resource_ptr<VulkanDescriptorSet> set) const;
-
 private:
+    bool hasExternalSampler(fvkmemory::resource_ptr<VulkanDescriptorSet> set);
+
     void updateSetAndLayout(fvkmemory::resource_ptr<VulkanDescriptorSet> set,
             fvkmemory::resource_ptr<VulkanDescriptorSetLayout> layout);
 

filament/backend/src/vulkan/VulkanHandles.h

@@ -308,10 +308,7 @@ struct VulkanRenderTarget : private HwRenderTarget, fvkmemory::Resource {
     // Creates a special "default" render target (i.e. associated with the swap chain)
     explicit VulkanRenderTarget();
 
-    VulkanRenderTarget(VulkanRenderTarget&& target)
-        : HwRenderTarget(0, 0),
-          mOffscreen(false),
-          mProtected(false) {
+    VulkanRenderTarget(VulkanRenderTarget&& target) {
         swap(std::move(target));
     }
 
@@ -368,8 +365,6 @@ struct VulkanRenderTarget : private HwRenderTarget, fvkmemory::Resource {
 
 private:
     void swap(VulkanRenderTarget&& target) {
-        std::swap(width, target.width);
-        std::swap(height, target.height);
         std::swap(mOffscreen, target.mOffscreen);
         std::swap(mProtected, target.mProtected);
         std::swap(mInfo, target.mInfo);

filament/backend/src/vulkan/VulkanQueryManager.cpp

@@ -46,13 +46,17 @@ fvkmemory::resource_ptr<VulkanTimerQuery> VulkanQueryManager::getNextQuery(
         return {};
     }
 
+    bool found = false;
     std::pair<uint32_t, uint32_t> queryIndices;
-    size_t const firstUnused = unused.firstSetBit();
-    {
+    unused.forEachSetBit([&](size_t index) {
+        if (found) {
+            return;
+        }
         std::unique_lock<utils::Mutex> lock(mMutex);
-        mUsed.set(firstUnused);
-        queryIndices = { firstUnused * 2, firstUnused * 2 + 1 };
-    }
+        mUsed.set(index);
+        found = true;
+        queryIndices = std::make_pair(index * 2, index * 2 + 1);
+    });
     return fvkmemory::resource_ptr<VulkanTimerQuery>::construct(resourceManager, queryIndices.first,
             queryIndices.second);
 }

filament/backend/src/vulkan/VulkanTexture.cpp

@@ -106,7 +106,7 @@ VkComponentMapping composeSwizzle(VkComponentMapping const& prev, VkComponentMap
         return { vals[0], vals[1], vals[2], vals[3] };
     };
 
-    // We make sure all identities are mapped into respective channels so that actual channel
+    // We make sure all all identities are mapped into respective channels so that actual channel
     // mapping will be passed onto the output.
     VkComponentMapping const prevExplicit = identityToChannel(prev);
     VkComponentMapping const nextExplicit = identityToChannel(next);

filament/backend/src/webgpu/WebGPUBlitter.cpp

@@ -58,10 +58,10 @@ namespace {
            areCopyCompatible(sourceFormat, destinationFormat) &&
            (sourceExtent.width == destinationExtent.width) &&
            (sourceExtent.height == destinationExtent.height) &&
-           ((sourceOrigin.x % sourceBlockSize.width) == 0) &&
-           ((sourceOrigin.y % sourceBlockSize.height) == 0) &&
-           ((destinationOrigin.x % destinationBlockSize.width) == 0) &&
-           ((destinationOrigin.y % destinationBlockSize.height) == 0);
+           ((sourceOrigin.x & sourceBlockSize.width) == 0) &&
+           ((sourceOrigin.y & sourceBlockSize.height) == 0) &&
+           ((destinationOrigin.x & destinationBlockSize.width) == 0) &&
+           ((destinationOrigin.y & destinationBlockSize.height) == 0);
 }
 
 struct BlitFragmentShaderArgs final {
@@ -206,20 +206,9 @@ constexpr std::string_view FRAGMENT_SHADER_SNIPPET_2D_INPUT_TEMPLATE{ R"(
 
 } // namespace
 
-// It can perform a direct memory copy if the formats are compatible and no scaling or format
-// conversion is needed. Otherwise, it uses a render pass with a custom shader to perform
-// the blit. This allows for scaling, format conversion, and resolving multisampled textures.
-// The blitter caches pipelines, layouts, and shaders to avoid recompilation.
 WebGPUBlitter::WebGPUBlitter(wgpu::Device const& device)
     : mDevice{ device } {}
 
-// Performs a blit operation from a source to a destination texture.
-// This function first checks if a direct texture-to-texture copy can be performed.
-// A direct copy is possible if the source and destination textures have the same sample count,
-// compatible formats, and the copy extents match. If a direct copy is not possible, this
-// function falls back to a render pass-based blit. The render pass uses a shader to read from
-// the source texture and write to the destination texture, allowing for format conversion,
-// scaling, and MSAA resolves.
 void WebGPUBlitter::blit(wgpu::Queue const& queue, wgpu::CommandEncoder const& commandEncoder,
         BlitArgs const& args) {
     // current assumptions/simplifications made in this implementation:
@@ -278,20 +267,22 @@ void WebGPUBlitter::blit(wgpu::Queue const& queue, wgpu::CommandEncoder const& c
                 args.destination.texture.GetUsage() & wgpu::TextureUsage::CopyDst)
                 << "destination texture usage doesn't have wgpu::TextureUsage::CopyDst";
 
-        copyByteByByte(commandEncoder, args);
+        copy(commandEncoder, args);
         return;
     }
+    // need to do a render pass blit...
 
     FILAMENT_CHECK_PRECONDITION(args.destination.texture.GetSampleCount() == 1)
             << "Blitting does not currently support writing to multisampled output textures.";
 
     FILAMENT_CHECK_PRECONDITION(args.source.texture.GetUsage() & wgpu::TextureUsage::TextureBinding)
-            << "Source texture usage doesn't have wgpu::TextureUsage::TextureBinding";
+            << "source texture usage doesn't have wgpu::TextureUsage::TextureBinding";
 
     FILAMENT_CHECK_PRECONDITION(
             args.destination.texture.GetUsage() & wgpu::TextureUsage::RenderAttachment)
-            << "Destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment";
+            << "destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment";
 
+    // create input/output texture views...
     const wgpu::TextureViewDimension sourceDimension{ args.source.texture.GetDimension() ==
                                                                       wgpu::TextureDimension::e3D
                                                               ? wgpu::TextureViewDimension::e3D
@@ -336,6 +327,7 @@ void WebGPUBlitter::blit(wgpu::Queue const& queue, wgpu::CommandEncoder const& c
             << destinationViewDescriptor.baseArrayLayer << " and mip level "
             << destinationViewDescriptor.baseMipLevel << " for render pass blit?";
 
+    // create uniform buffer (for shader args)...
     const BlitFragmentShaderArgs blitFragmentShaderArgs{
         .depthPlane = args.source.layerOrDepth,
         .scale = { static_cast<float>(args.source.extent.width) / args.destination.extent.width,
@@ -446,7 +438,7 @@ void WebGPUBlitter::blit(wgpu::Queue const& queue, wgpu::CommandEncoder const& c
     const wgpu::RenderPassEncoder renderPassEncoder{ commandEncoder.BeginRenderPass(
             &renderPassDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(renderPassEncoder)
-            << "Failed to create render pass encoder for blit.";
+            << "Failed to create render pass encoder for blit?";
     renderPassEncoder.SetPipeline(getOrCreateRenderPipeline(args.filter, sourceDimension,
             args.source.texture.GetSampleCount(), depthSource,
             args.destination.texture.GetFormat()));
@@ -456,7 +448,7 @@ void WebGPUBlitter::blit(wgpu::Queue const& queue, wgpu::CommandEncoder const& c
     renderPassEncoder.End();
 }
 
-void WebGPUBlitter::copyByteByByte(wgpu::CommandEncoder const& commandEncoder, BlitArgs const& args) {
+void WebGPUBlitter::copy(wgpu::CommandEncoder const& commandEncoder, BlitArgs const& args) {
     const wgpu::TexelCopyTextureInfo sourceCopyInfo{
         .texture = args.source.texture,
         .mipLevel = args.source.mipLevel,
@@ -513,7 +505,7 @@ void WebGPUBlitter::createSampler(const SamplerMagFilter filter) {
         .maxAnisotropy = 1, // should not matter, just being consistently defined
     };
     const wgpu::Sampler sampler{ mDevice.CreateSampler(&descriptor) };
-    FILAMENT_CHECK_POSTCONDITION(sampler) << "Failed to create blit sampler.";
+    FILAMENT_CHECK_POSTCONDITION(sampler) << "Failed to create blit sampler?";
     switch (filter) {
         case SamplerMagFilter::NEAREST:
             mNearestSampler = sampler;
@@ -524,8 +516,6 @@ void WebGPUBlitter::createSampler(const SamplerMagFilter filter) {
     }
 }
 
-// Caches and returns a render pipeline for a given blit configuration.
-// If a pipeline for the given configuration does not exist, it creates and caches one.
 wgpu::RenderPipeline const& WebGPUBlitter::getOrCreateRenderPipeline(
         const SamplerMagFilter filterType, const wgpu::TextureViewDimension sourceDimension,
         const uint32_t sourceSampleCount, const bool depthSource,
@@ -539,9 +529,6 @@ wgpu::RenderPipeline const& WebGPUBlitter::getOrCreateRenderPipeline(
     return mRenderPipelines[key];
 }
 
-// Creates a render pipeline for a blit operation.
-// The pipeline is configured with the appropriate vertex and fragment shaders,
-// pipeline layout, and render target state based on the blit parameters.
 wgpu::RenderPipeline WebGPUBlitter::createRenderPipeline(const SamplerMagFilter filterType,
         const wgpu::TextureViewDimension sourceDimension, const uint32_t sourceSampleCount,
         const bool depthSource, const wgpu::TextureFormat destinationTextureFormat) {
@@ -593,7 +580,7 @@ wgpu::RenderPipeline WebGPUBlitter::createRenderPipeline(const SamplerMagFilter
         .fragment = &fragmentState,
     };
     const wgpu::RenderPipeline pipeline{ mDevice.CreateRenderPipeline(&pipelineDescriptor) };
-    FILAMENT_CHECK_POSTCONDITION(pipeline) << "Failed to create pipeline for render pass blit.";
+    FILAMENT_CHECK_POSTCONDITION(pipeline) << "Failed to create pipeline for render pass blit?";
     return pipeline;
 }
 
@@ -608,8 +595,6 @@ size_t WebGPUBlitter::hashRenderPipelineKey(const SamplerMagFilter filterType,
     return seed;
 }
 
-// Caches and returns a pipeline layout for a given blit configuration.
-// If a layout for the given configuration does not exist, it creates and caches one.
 wgpu::PipelineLayout const& WebGPUBlitter::getOrCreatePipelineLayout(
         const SamplerMagFilter filterType, const wgpu::TextureViewDimension sourceDimension,
         const bool multisampledSource, const bool depthSource) {
@@ -622,8 +607,6 @@ wgpu::PipelineLayout const& WebGPUBlitter::getOrCreatePipelineLayout(
     return mPipelineLayouts[key];
 }
 
-// Creates a pipeline layout for a blit operation.
-// The layout defines the bind group layouts used by the blit pipeline.
 wgpu::PipelineLayout WebGPUBlitter::createPipelineLayout(const SamplerMagFilter filterType,
         const wgpu::TextureViewDimension sourceDimension, const bool multisampledSource,
         const bool depthSource) {
@@ -636,7 +619,7 @@ wgpu::PipelineLayout WebGPUBlitter::createPipelineLayout(const SamplerMagFilter
     const wgpu::PipelineLayout pipelineLayout{ mDevice.CreatePipelineLayout(
             &pipelineLayoutDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(pipelineLayout)
-            << "Failed to create pipeline layout for render pass blit.";
+            << "Failed to create pipeline layout for render pass blit?";
     return pipelineLayout;
 }
 
@@ -650,8 +633,6 @@ size_t WebGPUBlitter::hashPipelineLayoutKey(const SamplerMagFilter filterType,
     return seed;
 }
 
-// Caches and returns a bind group layout for a given blit configuration.
-// If a layout for the given configuration does not exist, it creates and caches one.
 wgpu::BindGroupLayout const& WebGPUBlitter::getOrCreateTextureBindGroupLayout(
         const SamplerMagFilter filterType, const wgpu::TextureViewDimension sourceDimension,
         const bool multisampledSource, const bool depthSource) {
@@ -664,9 +645,6 @@ wgpu::BindGroupLayout const& WebGPUBlitter::getOrCreateTextureBindGroupLayout(
     return mTextureBindGroupLayouts[key];
 }
 
-// Creates a bind group layout for the blit operation's texture resources.
-// This layout specifies the bindings for the source texture, a uniform buffer with blit parameters,
-// and optionally a sampler.
 wgpu::BindGroupLayout WebGPUBlitter::createTextureBindGroupLayout(const SamplerMagFilter filterType,
         const wgpu::TextureViewDimension sourceDimension, const bool multisampledSource,
         const bool depthSource) {
@@ -677,11 +655,11 @@ wgpu::BindGroupLayout WebGPUBlitter::createTextureBindGroupLayout(const SamplerM
             .texture = {
                 .sampleType = depthSource
                                       ? wgpu::TextureSampleType::Depth
-                                      : (multisampledSource
-                                                ? wgpu::TextureSampleType::UnfilterableFloat
-                                                : wgpu::TextureSampleType::Float), // only F32 scalar sample
-                                                                                   // type supported for now
-                                                                                   // (aside from depth)
+                                      : wgpu::TextureSampleType::Float, // only F32 scalar sample
+                                                                        // type supported for now
+                                                                        // (aside from depth)
+                                                                        // see assumptions listed
+                                                                        // in blit function
                 .viewDimension = sourceDimension,
                 .multisampled = multisampledSource,
             },
@@ -715,7 +693,7 @@ wgpu::BindGroupLayout WebGPUBlitter::createTextureBindGroupLayout(const SamplerM
     const wgpu::BindGroupLayout textureBindGroupLayout{ mDevice.CreateBindGroupLayout(
             &textureBindGroupLayoutDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(textureBindGroupLayout)
-            << "Failed to create texture bind group layout for render pass blit.";
+            << "Failed to create texture bind group layout for render pass blit?";
     return textureBindGroupLayout;
 }
 
@@ -729,8 +707,6 @@ size_t WebGPUBlitter::hashTextureBindGroupLayoutKey(const SamplerMagFilter filte
     return seed;
 }
 
-// Caches and returns a shader module for a given blit configuration.
-// If a shader module for the given configuration does not exist, it creates and caches one.
 wgpu::ShaderModule const& WebGPUBlitter::getOrCreateShaderModule(
         const wgpu::TextureViewDimension sourceDimension, const bool multisampledSource,
         const bool depthSource, const bool depthDestination) {
@@ -743,9 +719,6 @@ wgpu::ShaderModule const& WebGPUBlitter::getOrCreateShaderModule(
     return mShaderModules[key];
 }
 
-// Creates a shader module containing the vertex and fragment shaders for the blit operation.
-// The shader source is generated from a template, with placeholders filled in based on the
-// blit configuration (e.g., texture type, sample count).
 wgpu::ShaderModule WebGPUBlitter::createShaderModule(
         const wgpu::TextureViewDimension sourceDimension, const bool multisampledSource,
         const bool depthSource, const bool depthDestination) {
@@ -757,8 +730,9 @@ wgpu::ShaderModule WebGPUBlitter::createShaderModule(
             textureType = "texture_depth_2d";
         }
     } else {
+        // non-depth
         if (multisampledSource) {
-            textureType = "texture_multisampled_2d<f32>";
+            textureType = "texture_multisampled_2d";
         } else {
             if (sourceDimension == wgpu::TextureViewDimension::e3D) {
                 textureType = "texture_3d<f32>";
@@ -811,7 +785,7 @@ wgpu::ShaderModule WebGPUBlitter::createShaderModule(
     };
     const wgpu::ShaderModule shaderModule{ mDevice.CreateShaderModule(&shaderModuleDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(shaderModule)
-            << "Failed to create shader module for render pass blit.";
+            << "Failed to create shader module for render pass blit?";
     return shaderModule;
 }
 

filament/backend/src/webgpu/WebGPUBlitter.h

@@ -26,17 +26,6 @@
 
 namespace filament::backend {
 
-/**
-  * A blit essentially writes pixels to a region of one texture given a region of another texture.
-  * Such a process can be a byte-by-byte copy when possible, while other scenarios involve various
-  * transformations, such as pixel format conversion, image scaling, filtering/sampling
-  * (such as down sampling), or even multi-sample resolves. Some blits can be faster than others,
-  * depending on what is needed; for example, a byte-by-byte copy is generally faster than other
-  * operations.
-  *
-  * This class is responsible for performing blits throughout the lifecycle of its owner/caller,
-  * e.g. the WebGPU backend/driver, where it can leverage state to optimize such blit calls over time.
-  */
 class WebGPUBlitter final {
 public:
     struct BlitArgs final {
@@ -69,15 +58,10 @@ private:
     /**
      * ONLY should be called if canDoDirectCopy(...) is true (see it in WebGPUBlitter.cpp)
      */
-    void copyByteByByte(wgpu::CommandEncoder const&, BlitArgs const&);
+    void copy(wgpu::CommandEncoder const&, BlitArgs const&);
 
     void createSampler(SamplerMagFilter);
 
-    // The following methods are used to create and cache WebGPU objects.
-    // The pattern is to have a `getOrCreate...` method that looks up the object in a cache,
-    // and if it's not found, it calls a `create...` method to create it and then stores it in the
-    // cache. A `hash...` method is used to generate a key for the cache.
-
     [[nodiscard]] wgpu::RenderPipeline const& getOrCreateRenderPipeline(SamplerMagFilter,
             wgpu::TextureViewDimension sourceDimension, uint32_t sourceSampleCount,
             bool depthSource, wgpu::TextureFormat destinationTextureFormat);

filament/backend/src/webgpu/WebGPUBufferBase.cpp

@@ -33,9 +33,6 @@ namespace filament::backend {
 
 namespace {
 
-// Creates a wgpu::Buffer, ensuring its size is a multiple of FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS.
-// WebGPU's WriteBuffer requires the write size to be a multiple of 4. By ensuring the buffer
-// size is also a multiple of 4, we simplify the update logic.
 [[nodiscard]] wgpu::Buffer createBuffer(wgpu::Device const& device, const wgpu::BufferUsage usage,
         uint32_t size, const char* const label) {
     // Write size must be divisible by WEBGPU_BUFFER_SIZE_MODULUS (e.g. 4).
@@ -58,11 +55,6 @@ WebGPUBufferBase::WebGPUBufferBase(wgpu::Device const& device, const wgpu::Buffe
         const uint32_t size, char const* const label)
     : mBuffer{ createBuffer(device, usage, size, label) } {}
 
-// Updates the GPU buffer with data from a BufferDescriptor.
-// WebGPU requires that the size of the data written to a buffer is a multiple of 4.
-// This function handles cases where the buffer descriptor's size is not a multiple of 4
-// by writing the bulk of the data first, and then copying the remaining bytes into a
-// padded temporary chunk which is then written to the buffer.
 void WebGPUBufferBase::updateGPUBuffer(BufferDescriptor const& bufferDescriptor,
         const uint32_t byteOffset, wgpu::Queue const& queue) {
     FILAMENT_CHECK_PRECONDITION(bufferDescriptor.buffer)

filament/backend/src/webgpu/WebGPUBufferBase.h

@@ -28,10 +28,6 @@ namespace filament::backend {
 
 class BufferDescriptor;
 
-/**
-  * A base class for WebGPU buffer objects, providing common functionality for creating and
-  * updating GPU buffers.
-  */
 class WebGPUBufferBase /* intended to be extended */ {
 public:
     /**
@@ -50,8 +46,7 @@ protected:
 
 private:
     const wgpu::Buffer mBuffer;
-    // WebGPU requires that the source buffer of a writeBuffer call has a size that is a multiple
-    // of 4. This member is used to pad the data if the source size is not a multiple of 4.
+    // FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS (e.g. 4) bytes to hold any extra chunk we need.
     std::array<uint8_t, FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS> mRemainderChunk{};
 };
 

filament/backend/src/webgpu/WebGPUBufferObject.cpp

@@ -40,8 +40,6 @@ namespace {
 
 } // namespace
 
-// The usage flags are determined by the binding type, and always include CopyDst to allow for
-// updating the buffer.
 WebGPUBufferObject::WebGPUBufferObject(wgpu::Device const& device,
         const BufferObjectBinding bindingType, const uint32_t byteCount)
     : HwBufferObject{ byteCount },

filament/backend/src/webgpu/WebGPUBufferObject.h

@@ -31,12 +31,6 @@ namespace filament::backend {
 
 enum class BufferObjectBinding : uint8_t;
 
-/**
-  * A WebGPU-specific implementation of BufferObject.
-  * A BufferObject is a GPU buffer that can be used for a variety of purposes, such as storing
-  * uniform data, vertex attributes, or pixel data. The specific usage is determined by the
-  * BufferObjectBinding.
-  */
 class WebGPUBufferObject final : public HwBufferObject, public WebGPUBufferBase {
 public:
     WebGPUBufferObject(wgpu::Device const&, BufferObjectBinding, uint32_t byteCount);

filament/backend/src/webgpu/WebGPUConstants.h

@@ -21,7 +21,6 @@
 
 #include <cstdint>
 
-// WebGPU requires that the source buffer of a writeBuffer call has a size that is a multiple of 4.
 constexpr size_t FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS = 4;
 
 // FWGPU is short for Filament WebGPU
@@ -49,16 +48,8 @@ constexpr size_t FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS = 4;
 // textures, samplers, and buffers,...), which would otherwise spam the logs.
 #define FWGPU_DEBUG_BIND_GROUPS 0x00000020
 
-// Enables Android systrace
-#define FWGPU_DEBUG_SYSTRACE 0x00000040
-
-// Enable a minimal set of traces to assess the performance of the backend.
-// All other debug features must be disabled.
-#define FWGPU_DEBUG_PROFILING 0x00000080
-
 // Useful default combinations
-#define FWGPU_DEBUG_EVERYTHING (0xFFFFFFFF & ~FWGPU_DEBUG_PROFILING)
-#define FWGPU_DEBUG_PERFORMANCE FWGPU_DEBUG_SYSTRACE
+#define FWGPU_DEBUG_EVERYTHING     0xFFFFFFFF
 
 #if defined(FILAMENT_BACKEND_DEBUG_FLAG)
     #define FWGPU_DEBUG_FORWARDED_FLAG (FILAMENT_BACKEND_DEBUG_FLAG & FWGPU_DEBUG_EVERYTHING)
@@ -67,53 +58,13 @@ constexpr size_t FILAMENT_WEBGPU_BUFFER_SIZE_MODULUS = 4;
 #endif
 
 #ifndef NDEBUG
-    #define FWGPU_DEBUG_FLAGS (FWGPU_DEBUG_PERFORMANCE | FWGPU_DEBUG_FORWARDED_FLAG)
+    #define FWGPU_DEBUG_FLAGS FWGPU_DEBUG_FORWARDED_FLAG
 #else
-    #define FWGPU_DEBUG_FLAGS FWGPU_DEBUG_SYSTRACE
-#endif
-
-// Override the debug flags if we are forcing profiling mode
-#if defined(FILAMENT_FORCE_PROFILING_MODE)
-#undef FWGPU_DEBUG_FLAGS
-#define FWGPU_DEBUG_FLAGS (FWGPU_DEBUG_PROFILING)
+    #define FWGPU_DEBUG_FLAGS 0
 #endif
 
 #define FWGPU_ENABLED(flags) (((FWGPU_DEBUG_FLAGS) & (flags)) == (flags))
 
-#if FWGPU_ENABLED(FWGPU_DEBUG_PROFILING) && FWGPU_DEBUG_FLAGS != FWGPU_DEBUG_PROFILING
-#error PROFILING is exclusive; all other debug features must be disabled.
-#endif
-
-#if FWGPU_DEBUG_FLAGS == FWGPU_DEBUG_PROFILING
-
-    #ifndef NDEBUG
-        #error PROFILING is meaningless in DEBUG mode.
-    #endif
-
-    #define FWGPU_SYSTRACE_CONTEXT()
-    #define FWGPU_SYSTRACE_START(marker)
-    #define FWGPU_SYSTRACE_END()
-    #define FWGPU_SYSTRACE_SCOPE()
-    #define FWGPU_PROFILE_MARKER(marker)  PROFILE_SCOPE(marker)
-
-#elif FWGPU_ENABLED(FWGPU_DEBUG_SYSTRACE)
-
-    #include <private/utils/Tracing.h>
-
-    #define FWGPU_SYSTRACE_CONTEXT()      FILAMENT_TRACING_CONTEXT(FILAMENT_TRACING_CATEGORY_FILAMENT)
-    #define FWGPU_SYSTRACE_START(marker)  FILAMENT_TRACING_NAME_BEGIN(FILAMENT_TRACING_CATEGORY_FILAMENT, marker)
-    #define FWGPU_SYSTRACE_END()          FILAMENT_TRACING_NAME_END(FILAMENT_TRACING_CATEGORY_FILAMENT)
-    #define FWGPU_SYSTRACE_SCOPE()        FILAMENT_TRACING_CALL(FILAMENT_TRACING_CATEGORY_FILAMENT)
-    #define FWGPU_PROFILE_MARKER(marker)  FILAMENT_TRACING_CALL(FILAMENT_TRACING_CATEGORY_FILAMENT)
-
-#else
-    #define FWGPU_SYSTRACE_CONTEXT()
-    #define FWGPU_SYSTRACE_START(marker)
-    #define FWGPU_SYSTRACE_END()
-    #define FWGPU_SYSTRACE_SCOPE()
-    #define FWGPU_PROFILE_MARKER(marker)
-#endif
-
 #if FWGPU_ENABLED(FWGPU_DEBUG_FORCE_LOG_TO_I)
     #define FWGPU_LOGI LOG(INFO)
     #define FWGPU_LOGD FWGPU_LOGI

filament/backend/src/webgpu/WebGPUDescriptorSet.cpp

@@ -42,9 +42,6 @@ constexpr uint8_t INVALID_INDEX = MAX_DESCRIPTOR_COUNT + 1;
 
 } // namespace
 
-// The constructor initializes the descriptor set based on a given layout.
-// It pre-allocates space for the bind group entries and sets up a mapping from binding index
-// to the internal entry array index. This prepares the object to be populated with resources.
 WebGPUDescriptorSet::WebGPUDescriptorSet(wgpu::BindGroupLayout const& layout,
         std::vector<WebGPUDescriptorSetLayout::BindGroupEntryInfo> const& bindGroupEntries)
     : mLayout{ layout },
@@ -55,10 +52,11 @@ WebGPUDescriptorSet::WebGPUDescriptorSet(wgpu::BindGroupLayout const& layout,
     for (size_t i = 0; i < bindGroupEntries.size(); ++i) {
         mEntries[i].binding = bindGroupEntries[i].binding;
     }
-
-    // Create a mapping from binding index to entry index for efficient lookup.
+    // Establish the size of entries based on the layout. This should be reliable and efficient.
     assert_invariant(INVALID_INDEX > mEntryIndexByBinding.size());
-    std::fill(mEntryIndexByBinding.begin(), mEntryIndexByBinding.end(), INVALID_INDEX);
+    for (size_t i = 0; i < mEntryIndexByBinding.size(); i++) {
+        mEntryIndexByBinding[i] = INVALID_INDEX;
+    }
     for (size_t index = 0; index < mEntries.size(); index++) {
         wgpu::BindGroupEntry const& entry = mEntries[index];
         assert_invariant(entry.binding < mEntryIndexByBinding.size());
@@ -68,20 +66,19 @@ WebGPUDescriptorSet::WebGPUDescriptorSet(wgpu::BindGroupLayout const& layout,
 
 void WebGPUDescriptorSet::addEntry(const unsigned int index, wgpu::BindGroupEntry&& entry) {
     if (mBindGroup) {
-        // The bind group has already been created, so we can't add new entries.
-        // Filament guarantees that the descriptor set will not be modified after it has been locked.
+        // We will keep getting hits from future updates, but shouldn't do anything
+        // Filament guarantees this won't change after things have locked.
         return;
     }
     // TODO: Putting some level of trust that Filament is not going to reuse indexes or go past the
     // layout index for efficiency. Add guards if wrong.
     FILAMENT_CHECK_POSTCONDITION(index < mEntryIndexByBinding.size())
-            << "impossible/invalid index for a descriptor/binding (out of range or >= "
+            << "impossible/invalid index for a descriptor/binding (our of range or >= "
                "MAX_DESCRIPTOR_COUNT) "
             << index;
     uint8_t entryIndex = mEntryIndexByBinding[index];
     FILAMENT_CHECK_POSTCONDITION(entryIndex != INVALID_INDEX && entryIndex < mEntries.size())
-            << "Invalid binding index: " << index;
-
+            << "Invalid binding " << index;
     entry.binding = index;
     mEntries[entryIndex] = std::move(entry);
 }
@@ -97,9 +94,8 @@ wgpu::BindGroup WebGPUDescriptorSet::lockAndReturn(wgpu::Device const& device) {
         .entries = mEntries.data()
     };
     mBindGroup = device.CreateBindGroup(&descriptor);
-    FILAMENT_CHECK_POSTCONDITION(mBindGroup) << "Failed to create bind group.";
-
-    // Once the bind group is created, we can release the resources used to create it.
+    FILAMENT_CHECK_POSTCONDITION(mBindGroup) << "Failed to create bind group?";
+    // once we have created the bind group itself we should no longer need any other state
 #if FWGPU_ENABLED(FWGPU_DEBUG_BIND_GROUPS)
     FWGPU_LOGD << "WebGPUDescriptorSet (lockAndReturn):";
     FWGPU_LOGD << "  wgpu::BindGroupLayout handle: " << mLayout.Get();

filament/backend/src/webgpu/WebGPUDescriptorSet.h

@@ -32,46 +32,23 @@
 
 namespace filament::backend {
 
-/**
-  * A WebGPU-specific implementation of DescriptorSet.
-  * This class manages a collection of resource bindings (e.g., textures, buffers) that are used by
-  * a shader. It corresponds to a `wgpu::BindGroup` in WebGPU. The actual `wgpu::BindGroup` is
-  * created lazily when `lockAndReturn` is called, at which point the descriptor set becomes
-  * immutable.
-  */
 class WebGPUDescriptorSet final : public HwDescriptorSet {
 public:
     WebGPUDescriptorSet(wgpu::BindGroupLayout const&,
             std::vector<WebGPUDescriptorSetLayout::BindGroupEntryInfo> const&);
     ~WebGPUDescriptorSet();
 
-    /**
-     * Adds or updates a resource binding in the descriptor set.
-     * This method is called to populate the descriptor set with buffers, textures, and samplers.
-     * It is only valid to call this before the descriptor set is locked (i.e., before
-     * `lockAndReturn` is called).
-     */
     void addEntry(unsigned int index, wgpu::BindGroupEntry&& entry);
 
-    /**
-     * Finalizes the descriptor set by creating the wgpu::BindGroup. After this call, the
-     * descriptor set is considered "locked" and cannot be modified further.
-     */
     [[nodiscard]] wgpu::BindGroup lockAndReturn(wgpu::Device const&);
 
-    /**
-     * @return true if the descriptor set has been locked (i.e., the wgpu::BindGroup has been
-     * created).
-     */
     [[nodiscard]] bool getIsLocked() const { return mBindGroup != nullptr; }
 
     [[nodiscard]] size_t getEntitiesWithDynamicOffsetsCount() const {
         return mEntriesWithDynamicOffsetsCount;
     }
 
-    /**
-     * May be nullptr. Use lockAndReturn to create the bind group when appropriate
-     */
+    // May be nullptr. Use lockAndReturn to create the bind group when appropriate
     [[nodiscard]] wgpu::BindGroup const& getBindGroup() const { return mBindGroup; }
 
 #if FWGPU_ENABLED(FWGPU_DEBUG_BIND_GROUPS)
@@ -83,7 +60,6 @@ private:
     std::array<uint8_t, MAX_DESCRIPTOR_COUNT> mEntryIndexByBinding{};
     std::vector<wgpu::BindGroupEntry> mEntries;
     const size_t mEntriesWithDynamicOffsetsCount;
-    // This is created lazily when lockAndReturn is called.
     wgpu::BindGroup mBindGroup = nullptr;
 };
 

filament/backend/src/webgpu/WebGPUDescriptorSetLayout.cpp

@@ -40,9 +40,7 @@ namespace filament::backend {
 
 namespace {
 
-/**
- * Converts Filament shader stage flags to the corresponding WebGPU shader stage bitmask.
- */
+// Convert Filament Shader Stage Flags bitmask to webgpu equivalent
 [[nodiscard]] wgpu::ShaderStage filamentStageToWGPUStage(const ShaderStageFlags fFlags) {
     wgpu::ShaderStage retStages = wgpu::ShaderStage::None;
     if (any(ShaderStageFlags::VERTEX & fFlags)) {
@@ -72,6 +70,10 @@ WebGPUDescriptorSetLayout::WebGPUDescriptorSetLayout(DescriptorSetLayout const&
         baseLabel = temp->c_str();
     }
 
+    // TODO: layoutDescriptor has a "Label". Ideally we can get info on what this layout is for
+    // debugging. For now, hack an incrementing value.
+    static int layoutNum = 0;
+
     const unsigned int samplerCount =
             std::count_if(layout.bindings.begin(), layout.bindings.end(), [](auto& fEntry) {
                 return DescriptorSetLayoutBinding::isSampler(fEntry.type);
@@ -85,11 +87,6 @@ WebGPUDescriptorSetLayout::WebGPUDescriptorSetLayout(DescriptorSetLayout const&
         auto& wEntry = wEntries.emplace_back();
         auto& entryInfo = mBindGroupEntries.emplace_back();
         wEntry.visibility = filamentStageToWGPUStage(fEntry.stageFlags);
-
-        // In WebGPU, textures and samplers are separate bindings.
-        // We map the Filament binding index to two WebGPU binding indices:
-        // - texture: binding * 2
-        // - sampler: binding * 2 + 1
         wEntry.binding = fEntry.binding * 2;
         entryInfo.binding = wEntry.binding;
 
@@ -181,13 +178,11 @@ WebGPUDescriptorSetLayout::WebGPUDescriptorSetLayout(DescriptorSetLayout const&
         } else if (isCubeArrayTypeDescriptor(fEntry.type)) {
             wEntry.texture.viewDimension = wgpu::TextureViewDimension::CubeArray;
         }
+        // fEntry.count is unused currently
     }
-    // TODO: layoutDescriptor has a "Label". Ideally we can get info on what this layout is for
-    // debugging. For now, hack an incrementing value.
-    static int layoutNum = 0;
-    std::string label =  "layout_" + baseLabel + "_" + std::to_string(++layoutNum) ;
+    std::string label =  "layout_" + baseLabel + std::to_string(++layoutNum) ;
     const wgpu::BindGroupLayoutDescriptor layoutDescriptor{
-        .label{label.c_str()},
+        .label{label.c_str()}, // Use .c_str() if label needs to be const char*
         .entryCount = wEntries.size(),
         .entries = wEntries.data()
     };

filament/backend/src/webgpu/WebGPUDescriptorSetLayout.h

@@ -27,11 +27,6 @@
 
 namespace filament::backend {
 
-/**
- * A WebGPU implementation of the HwDescriptorSetLayout.
- * This class defines the layout of a descriptor set, which specifies the types and arrangement
- * of resources that can be bound to a pipeline.
- */
 class WebGPUDescriptorSetLayout final : public HwDescriptorSetLayout {
 public:
     struct BindGroupEntryInfo final {

filament/backend/src/webgpu/WebGPUDriver.cpp

@@ -32,12 +32,10 @@
 #include "WebGPUTextureHelpers.h"
 #include "WebGPUVertexBuffer.h"
 #include "WebGPUVertexBufferInfo.h"
-#include "webgpu/utils/AsyncTaskCounter.h"
 #include <backend/platforms/WebGPUPlatform.h>
 
 #include "CommandStreamDispatcher.h"
 #include "DriverBase.h"
-#include "SystraceProfile.h"
 #include "private/backend/Dispatcher.h"
 #include <backend/DriverEnums.h>
 #include <backend/Handle.h>
@@ -73,7 +71,7 @@ void setDefaultTargetFlags(WebGPURenderTarget& defaultRenderTarget,
     const wgpu::TextureFormat depthFormat{ swapChain.getDepthFormat() };
     TargetBufferFlags newTargetFlags{ filament::backend::TargetBufferFlags::NONE };
 
-    // Color is always assumed to be present in the default render target.
+    // Assuming Color always present in default render target.
     newTargetFlags |= filament::backend::TargetBufferFlags::COLOR;
     if (depthFormat != wgpu::TextureFormat::Undefined) {
         if (hasDepth(depthFormat)) {
@@ -88,12 +86,6 @@ void setDefaultTargetFlags(WebGPURenderTarget& defaultRenderTarget,
 
 } // namespace
 
-// The WebGPUDriver is the main entry point for the WebGPU backend. It is responsible for
-// creating and managing all WebGPU resources, and for submitting commands to the GPU.
-// It implements the Driver interface, which is the abstraction layer used by Filament's
-// renderer. The driver uses a handle-based system to manage resources, and it operates
-// asynchronously, with `create...S` methods for synchronous handle allocation and
-// `create...R` methods for resource creation on the backend thread.
 Driver* WebGPUDriver::create(WebGPUPlatform& platform, const Platform::DriverConfig& driverConfig) noexcept {
     constexpr size_t defaultSize = FILAMENT_WEBGPU_HANDLE_ARENA_SIZE_IN_MB * 1024U * 1024U;
     Platform::DriverConfig validConfig {driverConfig};
@@ -149,7 +141,6 @@ void WebGPUDriver::tick(int) {
 
 void WebGPUDriver::beginFrame(int64_t monotonic_clock_ns,
         int64_t refreshIntervalNs, uint32_t frameId) {
-    FWGPU_PROFILE_MARKER(PROFILE_NAME_BEGINFRAME);
 }
 
 void WebGPUDriver::setFrameScheduledCallback(Handle<HwSwapChain> sch,
@@ -166,11 +157,9 @@ void WebGPUDriver::setPresentationTime(int64_t monotonic_clock_ns) {
 }
 
 void WebGPUDriver::endFrame(const uint32_t /* frameId */) {
-    FWGPU_PROFILE_MARKER(PROFILE_NAME_ENDFRAME);
     mPipelineLayoutCache.onFrameEnd();
     mPipelineCache.onFrameEnd();
 
-    // Clear the currently bound descriptor sets at the end of the frame.
     for (size_t i = 0; i < MAX_DESCRIPTOR_SET_COUNT; i++) {
         mCurrentDescriptorSets[i] = {};
     }
@@ -180,11 +169,10 @@ void WebGPUDriver::endFrame(const uint32_t /* frameId */) {
 // A new command encoder is created fresh after it.
 // If a command encoder is not in flight (nullptr), then this is a noop.
 void WebGPUDriver::flush(int /* dummy */) {
-    FWGPU_SYSTRACE_SCOPE();
     if (mCommandEncoder == nullptr) {
         return;
     }
-    // Submit the commands recorded so far.
+    // submit the command buffer thus far...
     assert_invariant(mRenderPassEncoder == nullptr);
     wgpu::CommandBufferDescriptor commandBufferDescriptor{
         .label = "frame_command_after_flush",
@@ -201,24 +189,12 @@ void WebGPUDriver::flush(int /* dummy */) {
     assert_invariant(mCommandEncoder);
 }
 
-// Submits the currently recorded commands and waits for them to complete on the GPU.
-// This is a synchronous operation and should be used sparingly.
 void WebGPUDriver::finish(int /* dummy */) {
-    FWGPU_SYSTRACE_SCOPE();
-    mDevice.Tick();
-    mAdapter.GetInstance().ProcessEvents();
-    mReadPixelMapsCounter.waitForAllToFinish();
-    flush();
-
-    // Some applications/samples might call finish in the very beginning of an application before
-    // creating any resources/tasks to flush out the system on any pending work from previous runs.
-    // Currently, gltf_viewer sample on android exhibit this behavior.
-    // Bail out early to safeguard against those cases. Otherwise, OnSubmittedWorkDone will
-    // hang on work that was never submitted.
     if (mCommandEncoder == nullptr) {
         return;
     }
-    // Wait for all previously submitted work to finish.
+
+    flush();
     std::mutex syncPoint;
     std::condition_variable syncCondition;
     bool done = false;
@@ -238,43 +214,33 @@ void WebGPUDriver::finish(int /* dummy */) {
 }
 
 void WebGPUDriver::destroyRenderPrimitive(Handle<HwRenderPrimitive> rph) {
-    if (!rph) {
-        return;
+   if (rph) {
+        destructHandle<WebGPURenderPrimitive>(rph);
     }
-    FWGPU_SYSTRACE_SCOPE();
-    destructHandle<WebGPURenderPrimitive>(rph);
 }
 
 void WebGPUDriver::destroyVertexBufferInfo(Handle<HwVertexBufferInfo> vertexBufferInfoHandle) {
-    if (!vertexBufferInfoHandle) {
-        return;
+    if (vertexBufferInfoHandle) {
+        destructHandle<WebGPUVertexBufferInfo>(vertexBufferInfoHandle);
     }
-    FWGPU_SYSTRACE_SCOPE();
-    destructHandle<WebGPUVertexBufferInfo>(vertexBufferInfoHandle);
 }
 
 void WebGPUDriver::destroyVertexBuffer(Handle<HwVertexBuffer> vertexBufferHandle) {
-    if (!vertexBufferHandle) {
-        return;
+    if (vertexBufferHandle) {
+        destructHandle<WebGPUVertexBuffer>(vertexBufferHandle);
     }
-    FWGPU_SYSTRACE_SCOPE();
-    destructHandle<WebGPUVertexBuffer>(vertexBufferHandle);
 }
 
 void WebGPUDriver::destroyIndexBuffer(Handle<HwIndexBuffer> indexBufferHandle) {
-    if (!indexBufferHandle) {
-        return;
+    if (indexBufferHandle) {
+        destructHandle<WebGPUIndexBuffer>(indexBufferHandle);
     }
-    FWGPU_SYSTRACE_SCOPE();
-    destructHandle<WebGPUIndexBuffer>(indexBufferHandle);
 }
 
 void WebGPUDriver::destroyBufferObject(Handle<HwBufferObject> bufferObjectHandle) {
-    if (!bufferObjectHandle) {
-        return;
+    if (bufferObjectHandle) {
+        destructHandle<WebGPUBufferObject>(bufferObjectHandle);
     }
-    FWGPU_SYSTRACE_SCOPE();
-    destructHandle<WebGPUBufferObject>(bufferObjectHandle);
 }
 
 void WebGPUDriver::destroyTexture(Handle<HwTexture> textureHandle) {
@@ -298,8 +264,10 @@ void WebGPUDriver::destroyRenderTarget(Handle<HwRenderTarget> renderTargetHandle
         if (renderTarget == mCurrentRenderTarget) {
             mCurrentRenderTarget = nullptr;
         }
-        // The HwTexture handles stored within WGPURenderTarget are not owned by it,
-        // so they are not destroyed here. They are destroyed via WebGPUDriver::destroyTexture.
+        // WGPURenderTarget destructor is trivial.
+        // The HwTexture handles stored within WGPURenderTarget (via MRT, TargetBufferInfo)
+        // are not owned by WGPURenderTarget, so they are not destroyed here.
+        // They are destroyed via WebGPUDriver::destroyTexture.
         destructHandle<WebGPURenderTarget>(renderTargetHandle);
     }
 }
@@ -344,14 +312,6 @@ void WebGPUDriver::destroyDescriptorSet(Handle<HwDescriptorSet> descriptorSetHan
     }
 }
 
-// ------------------------------------------------------------------------------------------------
-// Synchronous handle creation methods (S-methods)
-//
-// These methods are called from the frontend thread to allocate a handle for a backend resource.
-// They are lightweight and do not perform any GPU operations. The actual resource creation
-// happens in the corresponding R-method on the backend thread.
-// ------------------------------------------------------------------------------------------------
-
 Handle<HwSwapChain> WebGPUDriver::createSwapChainS() noexcept {
     return allocHandle<WebGPUSwapChain>();
 }
@@ -436,17 +396,9 @@ Handle<HwTexture> WebGPUDriver::createTextureExternalImagePlaneS() noexcept {
     return allocHandle<WebGPUTexture>();
 }
 
-// ------------------------------------------------------------------------------------------------
-// Asynchronous resource creation methods (R-methods)
-//
-// These methods are called on the backend thread to create the actual GPU resources
-// associated with a handle that was previously allocated by an S-method.
-// ------------------------------------------------------------------------------------------------
-
-void WebGPUDriver::createSwapChainR(Handle<HwSwapChain> sch, void* nativeWindow,
-        const uint64_t flags) {
-    FWGPU_SYSTRACE_SCOPE();
+void WebGPUDriver::createSwapChainR(Handle<HwSwapChain> sch, void* nativeWindow, uint64_t flags) {
     mNativeWindow = nativeWindow;
+    assert_invariant(!mSwapChain);
     wgpu::Surface surface = mPlatform.createSurface(nativeWindow, flags);
 
     wgpu::Extent2D extent = mPlatform.getSurfaceExtent(mNativeWindow);
@@ -481,14 +433,12 @@ void WebGPUDriver::createSwapChainHeadlessR(Handle<HwSwapChain> sch, uint32_t wi
 
 void WebGPUDriver::createVertexBufferInfoR(Handle<HwVertexBufferInfo> vertexBufferInfoHandle,
         const uint8_t bufferCount, const uint8_t attributeCount, const AttributeArray attributes) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPUVertexBufferInfo>(vertexBufferInfoHandle, bufferCount, attributeCount,
             attributes, mDeviceLimits);
 }
 
 void WebGPUDriver::createVertexBufferR(Handle<HwVertexBuffer> vertexBufferHandle,
         const uint32_t vertexCount, Handle<HwVertexBufferInfo> vertexBufferInfoHandle) {
-    FWGPU_SYSTRACE_SCOPE();
     const auto vertexBufferInfo = handleCast<WebGPUVertexBufferInfo>(vertexBufferInfoHandle);
     constructHandle<WebGPUVertexBuffer>(vertexBufferHandle, vertexCount,
             vertexBufferInfo->bufferCount, vertexBufferInfoHandle);
@@ -496,14 +446,12 @@ void WebGPUDriver::createVertexBufferR(Handle<HwVertexBuffer> vertexBufferHandle
 
 void WebGPUDriver::createIndexBufferR(Handle<HwIndexBuffer> indexBufferHandle,
         const ElementType elementType, const uint32_t indexCount, const BufferUsage usage) {
-    FWGPU_SYSTRACE_SCOPE();
     const auto elementSize = static_cast<uint8_t>(getElementTypeSize(elementType));
     constructHandle<WebGPUIndexBuffer>(indexBufferHandle, mDevice, elementSize, indexCount);
 }
 
 void WebGPUDriver::createBufferObjectR(Handle<HwBufferObject> bufferObjectHandle,
         const uint32_t byteCount, const BufferObjectBinding bindingType, const BufferUsage usage) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPUBufferObject>(bufferObjectHandle, mDevice, bindingType, byteCount);
 }
 
@@ -511,7 +459,6 @@ void WebGPUDriver::createTextureR(Handle<HwTexture> textureHandle, const Sampler
         const uint8_t levels, const TextureFormat format, const uint8_t samples,
         const uint32_t width, const uint32_t height, const uint32_t depth,
         const TextureUsage usage) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPUTexture>(textureHandle, target, levels, format, samples, width, height,
             depth, usage, mDevice);
 }
@@ -527,32 +474,13 @@ void WebGPUDriver::createTextureViewSwizzleR(Handle<HwTexture> textureHandle,
         Handle<HwTexture> sourceTextureHandle, const backend::TextureSwizzle r,
         const backend::TextureSwizzle g, const backend::TextureSwizzle b,
         const backend::TextureSwizzle a) {
-
-    if (!isTextureSwizzleSupported()) {
-        FWGPU_LOGW << "WebGPUDriver::createTextureViewSwizzleR called while texture swizzling is "
-                      "not supported by the device/driver";
-    }
-
-    auto sourceTexture{ handleCast<WebGPUTexture>(sourceTextureHandle) };
-    assert_invariant(sourceTexture);
-
-    wgpu::TextureComponentSwizzle const nextSwizzle{
-        .r = toWGPUComponentSwizzle(r),
-        .g = toWGPUComponentSwizzle(g),
-        .b = toWGPUComponentSwizzle(b),
-        .a = toWGPUComponentSwizzle(a),
-    };
-
-    // The WebGPUTexture constructor for swizzled views will handle composing the swizzle and
-    // creating the new texture view.
-    constructHandle<WebGPUTexture>(textureHandle, sourceTexture, nextSwizzle);
+    PANIC_POSTCONDITION("Swizzle WebGPU Texture is not supported");
 }
 
 void WebGPUDriver::createTextureExternalImage2R(Handle<HwTexture> textureHandle,
         const backend::SamplerType target, const backend::TextureFormat format,
         const uint32_t width, const uint32_t height, const backend::TextureUsage usage,
         Platform::ExternalImageHandleRef externalImage) {
-    FWGPU_SYSTRACE_SCOPE();
     PANIC_POSTCONDITION("External WebGPU Texture is not supported");
 }
 
@@ -560,7 +488,6 @@ void WebGPUDriver::createTextureExternalImageR(Handle<HwTexture> textureHandle,
         const backend::SamplerType target, const backend::TextureFormat format,
         const uint32_t width, const uint32_t height, const backend::TextureUsage usage,
         void* externalImage) {
-    FWGPU_SYSTRACE_SCOPE();
     PANIC_POSTCONDITION("External WebGPU Texture is not supported");
 }
 
@@ -580,7 +507,6 @@ void WebGPUDriver::importTextureR(Handle<HwTexture> textureHandle, const intptr_
 void WebGPUDriver::createRenderPrimitiveR(Handle<HwRenderPrimitive> renderPrimitiveHandle,
         Handle<HwVertexBuffer> vertexBufferHandle, Handle<HwIndexBuffer> indexBufferHandle,
         const PrimitiveType primitiveType) {
-    FWGPU_SYSTRACE_SCOPE();
     assert_invariant(mDevice);
     const auto renderPrimitive = constructHandle<WebGPURenderPrimitive>(renderPrimitiveHandle);
     const auto vertexBuffer = handleCast<WebGPUVertexBuffer>(vertexBufferHandle);
@@ -591,7 +517,6 @@ void WebGPUDriver::createRenderPrimitiveR(Handle<HwRenderPrimitive> renderPrimit
 }
 
 void WebGPUDriver::createProgramR(Handle<HwProgram> programHandle, Program&& program) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPUProgram>(programHandle, mDevice, program);
 }
 
@@ -609,7 +534,6 @@ void WebGPUDriver::createRenderTargetR(Handle<HwRenderTarget> renderTargetHandle
         const TargetBufferFlags targetFlags, const uint32_t width, const uint32_t height,
         const uint8_t samples, const uint8_t layerCount, const MRT color,
         const TargetBufferInfo depth, const TargetBufferInfo stencil) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPURenderTarget>(
             renderTargetHandle, width, height, samples, layerCount, color, depth, stencil,
             targetFlags,
@@ -620,7 +544,7 @@ void WebGPUDriver::createRenderTargetR(Handle<HwRenderTarget> renderTargetHandle
 }
 
 void WebGPUDriver::createFenceR(Handle<HwFence> fenceHandle, const int /* dummy */) {
-    // The handle is constructed synchronously in createFenceS.
+    // the handle was already constructed in createFenceS
     const auto fence = handleCast<WebGPUFence>(fenceHandle);
     assert_invariant(mQueue);
     fence->addMarkerToQueueState(mQueue);
@@ -631,13 +555,11 @@ void WebGPUDriver::createTimerQueryR(Handle<HwTimerQuery> tqh, int) {}
 void WebGPUDriver::createDescriptorSetLayoutR(
         Handle<HwDescriptorSetLayout> descriptorSetLayoutHandle,
         backend::DescriptorSetLayout&& info) {
-    FWGPU_SYSTRACE_SCOPE();
     constructHandle<WebGPUDescriptorSetLayout>(descriptorSetLayoutHandle, std::move(info), mDevice);
 }
 
 void WebGPUDriver::createDescriptorSetR(Handle<HwDescriptorSet> descriptorSetHandle,
         Handle<HwDescriptorSetLayout> descriptorSetLayoutHandle) {
-    FWGPU_SYSTRACE_SCOPE();
     auto layout = handleCast<WebGPUDescriptorSetLayout>(descriptorSetLayoutHandle);
     constructHandle<WebGPUDescriptorSet>(descriptorSetHandle, layout->getLayout(),
             layout->getBindGroupEntries());
@@ -692,12 +614,14 @@ FenceStatus WebGPUDriver::getFenceStatus(Handle<HwFence> fenceHandle) {
     return fence->getStatus();
 }
 
+// We create all textures using VK_IMAGE_TILING_OPTIMAL, so our definition of "supported" is that
+// the GPU supports the given texture format with non-zero optimal tiling features.
 bool WebGPUDriver::isTextureFormatSupported(const TextureFormat format) {
     return toWGPUTextureFormat(format) != wgpu::TextureFormat::Undefined;
 }
 
 bool WebGPUDriver::isTextureSwizzleSupported() {
-    return mDevice.HasFeature(wgpu::FeatureName::TextureComponentSwizzle);
+    return false;
 }
 
 bool WebGPUDriver::isTextureFormatMipmappable(const TextureFormat format) {
@@ -872,10 +796,6 @@ void WebGPUDriver::setVertexBufferObject(Handle<HwVertexBuffer> vertexBufferHand
     vertexBuffer->getBuffers()[index] = bufferObject->getBuffer();
 }
 
-// Updates a 3D texture region with pixel data from a buffer.
-// This function handles various pixel formats and performs a blit operation if the source
-// format needs to be converted to the destination format. For compressed textures, it ensures
-// that the offsets are aligned to the block dimensions.
 void WebGPUDriver::update3DImage(Handle<HwTexture> textureHandle, const uint32_t level,
         const uint32_t xoffset, const uint32_t yoffset, const uint32_t zoffset,
         const uint32_t width, const uint32_t height, const uint32_t depth,
@@ -892,15 +812,15 @@ void WebGPUDriver::update3DImage(Handle<HwTexture> textureHandle, const uint32_t
     const auto texture{ handleCast<WebGPUTexture>(textureHandle) };
     FILAMENT_CHECK_PRECONDITION((texture->getTexture().GetWidth() + xoffset) >= width)
             << "Blitting requires the destination region to have enough width ("
-            << texture->getTexture().GetWidth() << ") to accommodate the requested " << width
+            << texture->getTexture().GetWidth() << " to accommodate the requested " << width
             << " width to write/blit (accounting for xoffset, which is " << xoffset << ").";
     FILAMENT_CHECK_PRECONDITION((texture->getTexture().GetHeight() + yoffset) >= height)
             << "Blitting requires the destination region to have enough height ("
-            << texture->getTexture().GetHeight() << ") to accommodate the requested " << height
+            << texture->getTexture().GetHeight() << " to accommodate the requested " << height
             << " height to write/blit (accounting for yoffset, which is " << yoffset << ").";
     FILAMENT_CHECK_PRECONDITION((texture->getTexture().GetDepthOrArrayLayers() + zoffset) >= depth)
             << "Blitting requires the destination region to have enough depth/arrayLayers ("
-            << texture->getTexture().GetDepthOrArrayLayers() << ") to accommodate the requested "
+            << texture->getTexture().GetDepthOrArrayLayers() << " to accommodate the requested "
             << depth << " depth to write/blit (accounting for zoffset, which is " << zoffset << ").";
 
     // TODO: Writing to a depth texture is illegal and errors. I'm not sure why Filament is trying
@@ -913,8 +833,10 @@ void WebGPUDriver::update3DImage(Handle<HwTexture> textureHandle, const uint32_t
             inputData->type) };
     const wgpu::TextureFormat outputLinearFormat{ toLinearFormat(
             texture->getTexture().GetFormat()) };
-
-    const bool doBlit{ conversionNecessary(inputPixelFormat, outputLinearFormat, inputData->type) };
+    const bool conversionNecessary{
+        inputPixelFormat != outputLinearFormat && inputData->type != PixelDataType::COMPRESSED
+    }; // compressed formats should never need conversion
+    const bool doBlit{ conversionNecessary };
 #if FWGPU_ENABLED(FWGPU_DEBUG_UPDATE_IMAGE)
     if (texture->width > 1000 && texture->height > 500) {
         FWGPU_LOGD << "Update3DImage(..., level=" << level << ", xoffset=" << xoffset
@@ -979,7 +901,7 @@ void WebGPUDriver::update3DImage(Handle<HwTexture> textureHandle, const uint32_t
                 << "Failed to create staging input texture for blit?";
         const auto copyInfo{ wgpu::TexelCopyTextureInfo{
             .texture = stagingTexture,
-            .mipLevel = 0,
+            .mipLevel = level,
             .origin = { .x = 0, .y = 0, .z = 0 },
             .aspect = texture->getAspect(),
         } };
@@ -1034,8 +956,9 @@ void WebGPUDriver::update3DImage(Handle<HwTexture> textureHandle, const uint32_t
             mQueue.Submit(1, &blitCommand);
             mCommandEncoder = nullptr;
         }
+        stagingTexture.Destroy();
     } else {
-        // Direct copy without a blit.
+        // not doing blit (copy byte-by-byte)...
         const auto copyInfo { wgpu::TexelCopyTextureInfo{
             .texture = texture->getTexture(),
             .mipLevel = level,
@@ -1147,15 +1070,10 @@ void WebGPUDriver::compilePrograms(CompilerPriorityQueue priority,
     }
 }
 
-// Begins a render pass. This sets up the render pass descriptor with the appropriate
-// color and depth/stencil attachments, based on whether the target is the default
-// swap chain render target or a custom one. It also handles MSAA by setting up
-// resolve targets if necessary.
 void WebGPUDriver::beginRenderPass(Handle<HwRenderTarget> renderTargetHandle,
         RenderPassParams const& params) {
-    FWGPU_SYSTRACE_SCOPE();
     if (!mCommandEncoder) {
-        wgpu::CommandEncoderDescriptor commandEncoderDescriptor{ .label = "frame_command_encoder" };
+        wgpu::CommandEncoderDescriptor commandEncoderDescriptor = { .label = "frame_command_encoder" };
         mCommandEncoder = mDevice.CreateCommandEncoder(&commandEncoderDescriptor);
     }
     assert_invariant(mCommandEncoder);
@@ -1307,7 +1225,6 @@ void WebGPUDriver::beginRenderPass(Handle<HwRenderTarget> renderTargetHandle,
 }
 
 void WebGPUDriver::endRenderPass(int /* dummy */) {
-    FWGPU_SYSTRACE_SCOPE();
     mRenderPassEncoder.End();
     mRenderPassEncoder = nullptr;
 }
@@ -1316,12 +1233,10 @@ void WebGPUDriver::nextSubpass(int) {
     //todo
 }
 
-void WebGPUDriver::makeCurrent(Handle<HwSwapChain> drawSwapChain,
-        Handle<HwSwapChain> readSwapChain) {
-    FWGPU_SYSTRACE_SCOPE();
-    ASSERT_PRECONDITION_NON_FATAL(drawSwapChain == readSwapChain,
+void WebGPUDriver::makeCurrent(Handle<HwSwapChain> drawSch, Handle<HwSwapChain> readSch) {
+    ASSERT_PRECONDITION_NON_FATAL(drawSch == readSch,
             "WebGPU driver does not support distinct draw/read swap chains.");
-    auto swapChain = handleCast<WebGPUSwapChain>(drawSwapChain);
+    auto swapChain = handleCast<WebGPUSwapChain>(drawSch);
     mSwapChain = swapChain;
     assert_invariant(mSwapChain);
 
@@ -1334,16 +1249,13 @@ void WebGPUDriver::makeCurrent(Handle<HwSwapChain> drawSwapChain,
     assert_invariant(mTextureView);
 
     if (!mCommandEncoder) {
-        wgpu::CommandEncoderDescriptor commandEncoderDescriptor = {
-            .label = "frame_command_encoder"
-        };
+        wgpu::CommandEncoderDescriptor commandEncoderDescriptor = { .label = "frame_command_encoder" };
         mCommandEncoder = mDevice.CreateCommandEncoder(&commandEncoderDescriptor);
     }
     assert_invariant(mCommandEncoder);
 }
 
-void WebGPUDriver::commit(Handle<HwSwapChain> swapChainHandle) {
-    FWGPU_SYSTRACE_SCOPE();
+void WebGPUDriver::commit(Handle<HwSwapChain> sch) {
     if (UTILS_VERY_UNLIKELY(!mCommandEncoder)) {
         // nothing to submit
         if (UTILS_LIKELY(mSwapChain)) {
@@ -1394,41 +1306,34 @@ void WebGPUDriver::insertEventMarker(char const* string) {
 }
 
 void WebGPUDriver::pushGroupMarker(char const* string) {
-    FWGPU_SYSTRACE_CONTEXT();
-    FWGPU_SYSTRACE_START(string);
-}
-
-void WebGPUDriver::popGroupMarker(int /* dummy */) {
-    FWGPU_SYSTRACE_CONTEXT();
-    FWGPU_SYSTRACE_END();
-}
-
-void WebGPUDriver::startCapture(int /* dummy */) {
     //todo
 }
 
-void WebGPUDriver::stopCapture(int /* dummy */) {
+void WebGPUDriver::popGroupMarker(int) {
+    //todo
+}
+
+void WebGPUDriver::startCapture(int) {
+    //todo
+}
+
+void WebGPUDriver::stopCapture(int) {
     //todo
 }
 
-// Reads a block of pixels from a render target into a buffer.
-// This function is asynchronous. It copies the pixel data to a staging buffer on the GPU,
-// and then maps the buffer for reading on the CPU. The provided callback is invoked when the
-// data is ready. This function also handles the 256-byte row alignment requirement for
-// buffer-to-texture copies in WebGPU.
 void WebGPUDriver::readPixels(Handle<HwRenderTarget> sourceRenderTargetHandle, const uint32_t x,
         const uint32_t y, const uint32_t width, const uint32_t height,
         PixelBufferDescriptor&& pixelBufferDescriptor) {
-    const auto srcTarget{ handleCast<WebGPURenderTarget>(sourceRenderTargetHandle) };
+    auto srcTarget = handleCast<WebGPURenderTarget>(sourceRenderTargetHandle);
     assert_invariant(srcTarget);
 
-    wgpu::Texture srcTexture{ nullptr };
+    wgpu::Texture srcTexture = nullptr;
     if (srcTarget->isDefaultRenderTarget()) {
         assert_invariant(mSwapChain);
         srcTexture = mSwapChain->getCurrentTexture();
     } else {
         // Handle custom render targets. Read from the first color attachment.
-        const auto& colorAttachmentInfos{ srcTarget->getColorAttachmentInfos() };
+        const auto& colorAttachmentInfos = srcTarget->getColorAttachmentInfos();
         // TODO we are currently assuming the first attachment is the desired texture.
         if (colorAttachmentInfos[0].handle) {
             auto texture = handleCast<WebGPUTexture>(colorAttachmentInfos[0].handle);
@@ -1443,172 +1348,116 @@ void WebGPUDriver::readPixels(Handle<HwRenderTarget> sourceRenderTargetHandle, c
         scheduleDestroy(std::move(pixelBufferDescriptor));
         return;
     }
-    const uint32_t srcWidth {srcTexture.GetWidth()};
-    const uint32_t srcHeight{srcTexture.GetHeight()};
+    const uint32_t srcWidth = srcTexture.GetWidth();
+    const uint32_t srcHeight = srcTexture.GetHeight();
 
     // Clamp read region to texture bounds
-    if (UTILS_UNLIKELY(x >= srcWidth || y >= srcHeight)) {
+    if (x >= srcWidth || y >= srcHeight) {
         scheduleDestroy(std::move(pixelBufferDescriptor));
         return;
     }
-    auto actualWidth{ std::min(width, srcWidth - x) };
-    auto actualHeight{ std::min(height, srcHeight - y)};
+    auto actualWidth = std::min(width, srcWidth - x);
+    auto actualHeight = std::min(height, srcHeight - y);
     if (UTILS_UNLIKELY(actualWidth == 0 || actualHeight == 0)) {
         scheduleDestroy(std::move(pixelBufferDescriptor));
         return;
     }
 
-    if (!mCommandEncoder) {
-        const wgpu::CommandEncoderDescriptor commandEncoderDescriptor{
-            .label = "read_pixels_command",
-        };
-        mCommandEncoder = mDevice.CreateCommandEncoder(&commandEncoderDescriptor);
-        FILAMENT_CHECK_POSTCONDITION(mCommandEncoder)
-                << "Failed to create command encoder for readPixels?";
-    }
+    // Once we're ready to read the pixels, we need to flush all the previous work of this frame.
+    // This ensures that the readPixels will be ordered after all the draws.
+    flush();
 
-    const wgpu::TextureFormat srcFormat{srcTexture.GetFormat()};
-    const wgpu::TextureFormat dstFormat{toWebGPUFormat(pixelBufferDescriptor.format,
-                                                       pixelBufferDescriptor.type)};
-    wgpu::Texture textureToReadFrom{srcTexture};
-    wgpu::Texture stagingTexture{ nullptr };
-    uint32_t readX{x};
-    uint32_t readY{y};
+    const size_t dstBytesPerPixel = PixelBufferDescriptor::computePixelSize(
+            pixelBufferDescriptor.format, pixelBufferDescriptor.type);
+    const size_t srcBytesPerPixel = getWGPUTextureFormatPixelSize(srcTexture.GetFormat());
 
-    // If the source format is different from the destination (e.g. BGRA vs RGBA),
-    // we need to perform a conversion using an intermediate blit.
-    if (conversionNecessary(srcFormat, dstFormat, pixelBufferDescriptor.type)) {
-        const wgpu::TextureDescriptor stagingDescriptor{
-                .label = "readpixels_staging_texture",
-                .usage = wgpu::TextureUsage::CopySrc | wgpu::TextureUsage::RenderAttachment,
-                .dimension = wgpu::TextureDimension::e2D, // are we sure we can assume this? thus far we have
-                .size = {
-                        .width = actualWidth,
-                        .height = actualHeight,
-                        .depthOrArrayLayers = 1,
-                },
-                .format = dstFormat,
-                .mipLevelCount = 1,
-                .sampleCount = srcTexture.GetSampleCount(),
-        };
-        stagingTexture = mDevice.CreateTexture(&stagingDescriptor);
-        assert_invariant(stagingTexture);
-        textureToReadFrom = stagingTexture;
-        const WebGPUBlitter::BlitArgs blitArgs{
-                .source = {
-                        .texture = srcTexture,
-                        .origin = {.x = x, .y = y,},
-                        .extent = {
-                                .width = actualWidth,
-                                .height = actualHeight,
-                        },
-                },
-                .destination = {
-                        .texture = stagingTexture,
-                        .origin = {.x = 0, .y = 0},
-                        .extent = {
-                                .width = actualWidth,
-                                .height = actualHeight,
-                        },
-                },
-                .filter = SamplerMagFilter::NEAREST, // Use NEAREST for a 1:1 copy
-        };
-        mBlitter.blit(mQueue, mCommandEncoder, blitArgs);
+    FILAMENT_CHECK_PRECONDITION(dstBytesPerPixel == srcBytesPerPixel && dstBytesPerPixel > 0)
+            << "Source texture pixel size (" << srcBytesPerPixel
+            << ") does not match destination pixel buffer pixel size (" << dstBytesPerPixel
+            << "), or the format is not supported for readPixels.";
 
-        // The subsequent read will be from the top-left of the new intermediate texture.
-        readX = 0;
-        readY = 0;
-    }
+    // Create a staging buffer to copy the texture to. WebGPU requires 256 byte alignment
+    const size_t bytesPerPixel = dstBytesPerPixel;
+    const size_t unpaddedBytesPerRow = actualWidth * bytesPerPixel;
+    const size_t alignment = 256;
+    const size_t paddedBytesPerRow = (unpaddedBytesPerRow + alignment - 1) & ~(alignment - 1);
 
-    // Create a staging buffer to copy the texture to. WebGPU requires 256 byte alignment for buffer-to-texture copies.
-    const size_t bytesPerPixel{ PixelBufferDescriptor::computePixelSize(
-            pixelBufferDescriptor.format, pixelBufferDescriptor.type) };
-    const size_t unpaddedBytesPerRow{ actualWidth * bytesPerPixel };
-    const size_t alignment{ 256 };
-    const size_t paddedBytesPerRow = { (unpaddedBytesPerRow + alignment - 1) & ~(alignment - 1) };
-
-    const size_t bufferSize{ paddedBytesPerRow * actualHeight };
-    const wgpu::BufferDescriptor bufferDesc{
-        .usage = wgpu::BufferUsage::CopyDst | wgpu::BufferUsage::MapRead,
-        .size = bufferSize,
-    };
-
-    wgpu::Buffer stagingBuffer{ mDevice.CreateBuffer(&bufferDesc) };
+    size_t bufferSize = paddedBytesPerRow * height;
+    wgpu::BufferDescriptor bufferDesc;
+    bufferDesc.size = bufferSize;
+    bufferDesc.usage = wgpu::BufferUsage::CopyDst | wgpu::BufferUsage::MapRead;
+    wgpu::Buffer stagingBuffer = mDevice.CreateBuffer(&bufferDesc);
     assert_invariant(stagingBuffer);
 
-    // WebGPU's texture coordinates for copies are top-left, but Filament's y-coordinate is
-    // bottom-left. We must flip the y-coordinate relative to the texture we are reading from.
-    const uint32_t textureHeight{ textureToReadFrom.GetHeight() };
-    const uint32_t flippedY{ textureHeight - readY - actualHeight };
+    wgpu::CommandEncoder encoder = mDevice.CreateCommandEncoder();
+    assert_invariant(encoder);
 
-    const wgpu::TexelCopyTextureInfo source{
-            .texture = textureToReadFrom, // Read from the original or intermediate texture
-            .mipLevel = 0,
-            .origin = {.x = readX, .y = flippedY, .z = 0,},
+    // WebGPU flips the y axis like Metal does
+    const uint32_t flippedY = srcHeight - y - height;
+    wgpu::TexelCopyTextureInfo source{
+        .texture = srcTexture,
+        .mipLevel = 0,
+        .origin = { .x = x, .y = flippedY, .z = 0 },
     };
-    const wgpu::TexelCopyBufferInfo destination{
-            .layout = {
-                    .offset = 0,
-                    .bytesPerRow = static_cast<uint32_t>(paddedBytesPerRow),
-                    .rowsPerImage = actualHeight,
-            },
-            .buffer = stagingBuffer
+    wgpu::TexelCopyBufferInfo destination {
+        .layout = {
+            .offset = 0,
+            .bytesPerRow = static_cast<uint32_t>(paddedBytesPerRow),
+            .rowsPerImage = actualHeight,
+        },
+        .buffer = stagingBuffer
     };
-    const wgpu::Extent3D copySize{
-        .width = actualWidth,
+    wgpu::Extent3D copySize{ .width = actualWidth,
         .height = actualHeight,
-        .depthOrArrayLayers = 1,
-    };
-    mCommandEncoder.CopyTextureToBuffer(&source, &destination, &copySize);
-    mCommandBuffer = mCommandEncoder.Finish();
-    assert_invariant(mCommandBuffer);
-    mCommandEncoder = nullptr;
-    mQueue.Submit(1, &mCommandBuffer);
-    mCommandBuffer = nullptr;
+        .depthOrArrayLayers = 1 };
+    // TODO consider removing encoder finish + submit once we have test coverage and can
+    //      safely verify doing so will not result in strange errors. We don't think this
+    //      submit is necessary, as the encoder should get finished and the command
+    //      submitted in commit(), finish(), and/or flush() in this frame
+    //      or the next anyway. And, the read/copy bits will be encoded before other things.
+    //      Thus, not submitting here _seems_ ok. Nonetheless, doing the submit here likely
+    //      represents unnecessary inefficiency and higher bandwidth
+    //      between the CPU and GPU, but we are worried about functional correctness at this
+    //      stage.
+    encoder.CopyTextureToBuffer(&source, &destination, &copySize);
+    wgpu::CommandBuffer commandBuffer = encoder.Finish();
+    mQueue.Submit(1, &commandBuffer);
 
     // Map the buffer to read the data
-    struct UserData final {
+    struct UserData {
         PixelBufferDescriptor pixelBufferDescriptor;
         wgpu::Buffer buffer;
         size_t unpaddedBytesPerRow;
         size_t paddedBytesPerRow;
         uint32_t height;
-        WebGPUDriver *driver;
+        WebGPUDriver* driver;
     };
     auto userData = std::make_unique<UserData>(UserData{
-            .pixelBufferDescriptor = std::move(pixelBufferDescriptor),
-            .buffer = std::move(stagingBuffer),
-            .unpaddedBytesPerRow = unpaddedBytesPerRow,
-            .paddedBytesPerRow = paddedBytesPerRow,
-            .height = actualHeight,
-            .driver = this,
+        .pixelBufferDescriptor = std::move(pixelBufferDescriptor),
+        .buffer = std::move(stagingBuffer),
+        .unpaddedBytesPerRow = unpaddedBytesPerRow,
+        .paddedBytesPerRow = paddedBytesPerRow,
+        .height = height,
+        .driver = this,
     });
 
-    mReadPixelMapsCounter.startTask();
     userData->buffer.MapAsync(
             wgpu::MapMode::Read, 0, bufferSize, wgpu::CallbackMode::AllowSpontaneous,
             [](wgpu::MapAsyncStatus status, const char* message, UserData* userdata) {
                 std::unique_ptr<UserData> data(static_cast<UserData*>(userdata));
                 if (UTILS_LIKELY(status == wgpu::MapAsyncStatus::Success)) {
-                    const char *src {static_cast<const char *>(
-                            data->buffer.GetConstMappedRange(0, data->buffer.GetSize()))};
-                    char* dst{ static_cast<char*>(data->pixelBufferDescriptor.buffer) };
-
-                    // If padding was added for alignment, we need to copy row by row.
-                    if (data->paddedBytesPerRow == data->unpaddedBytesPerRow) {
-                        memcpy(dst, src, data->unpaddedBytesPerRow * data->height);
-                    } else {
-                        for (uint32_t i{ 0 }; i < data->height; ++i) {
-                            memcpy(dst + i * data->unpaddedBytesPerRow,
-                                    src + i * data->paddedBytesPerRow, data->unpaddedBytesPerRow);
-                        }
+                    const char* src = static_cast<const char*>(
+                            data->buffer.GetConstMappedRange(0, data->buffer.GetSize()));
+                    char* dst = static_cast<char*>(data->pixelBufferDescriptor.buffer);
+                    for (uint32_t i = 0; i < data->height; ++i) {
+                        memcpy(dst + i * data->unpaddedBytesPerRow,
+                                src + i * data->paddedBytesPerRow, data->unpaddedBytesPerRow);
                     }
                     data->buffer.Unmap();
                 } else {
                     FWGPU_LOGE << "Failed to map staging buffer for readPixels: " << message;
                 }
                 data->driver->scheduleDestroy(std::move(data->pixelBufferDescriptor));
-                data->driver->mReadPixelMapsCounter.finishTask();
             },
             userData.release());
 }
@@ -1623,99 +1472,14 @@ void WebGPUDriver::blitDEPRECATED(TargetBufferFlags buffers,
         Handle<HwRenderTarget> destinationRenderTargetHandle, const Viewport destinationViewport,
         Handle<HwRenderTarget> sourceRenderTargetHandle, const Viewport sourceViewport,
         const SamplerMagFilter filter) {
-    FWGPU_SYSTRACE_SCOPE();
-    auto const sourceTarget{ handleCast<WebGPURenderTarget>(sourceRenderTargetHandle) };
-    auto const destinationTarget{ handleCast<WebGPURenderTarget>(destinationRenderTargetHandle) };
-    assert_invariant(sourceTarget && destinationTarget);
-
-    FILAMENT_CHECK_PRECONDITION(buffers == TargetBufferFlags::COLOR0)
-            << "blitDEPRECATED only supports COLOR0";
-
-    FILAMENT_CHECK_PRECONDITION(sourceViewport.left >= 0 && sourceViewport.bottom >= 0 &&
-                                destinationViewport.left >= 0 && destinationViewport.bottom >= 0)
-            << "Source and destination viewports must be positive.";
-
-    // We always blit from/to the COLOR0 attachment.
-    auto sourceAttachment {sourceTarget->getColorAttachmentInfos()[0]};
-    Handle<HwTexture> sourceTextureHandle{ sourceAttachment.handle };
-    auto destinationAttachment {destinationTarget->getColorAttachmentInfos()[0]};
-    Handle<HwTexture> destinationTextureHandle{
-        destinationAttachment.handle
-    };
-
-    if (UTILS_UNLIKELY(!sourceTextureHandle || !destinationTextureHandle)) {
-        FWGPU_LOGE << "blitDEPRECATED could not find a valid color attachment to blit.";
-        return;
-    }
-
-    // WebGPU's texture coordinates are top-left, while Filament's are bottom-left.
-    // We need to flip the y-coordinate for both source and destination.
-    auto const sourceTexture{ handleCast<WebGPUTexture>(sourceTextureHandle) };
-    auto const destinationTexture{ handleCast<WebGPUTexture>(destinationTextureHandle) };
-
-    const uint32_t sourceTextureHeight{ sourceTexture->height };
-    const uint32_t flippedSourceY{ sourceTextureHeight - sourceViewport.bottom -
-                                   sourceViewport.height };
-
-    const uint32_t destinationTextureHeight{ destinationTexture->height };
-    const uint32_t flippedDestinationY{ destinationTextureHeight - destinationViewport.bottom -
-                                        destinationViewport.height };
-
-    const wgpu::Origin2D sourceOrigin{ static_cast<uint32_t>(sourceViewport.left), flippedSourceY };
-    const wgpu::Origin2D destinationOrigin{ static_cast<uint32_t>(destinationViewport.left),
-        flippedDestinationY };
-
-    const wgpu::Extent2D sourceSize{ static_cast<uint32_t>(sourceViewport.width),
-        static_cast<uint32_t>(sourceViewport.height) };
-    const wgpu::Extent2D destinationSize{ static_cast<uint32_t>(destinationViewport.width),
-        static_cast<uint32_t>(destinationViewport.height) };
-
-    bool reusedCommandEncoder{ true };
-    if (mCommandEncoder) {
-        flush();
-    } else {
-        reusedCommandEncoder = false;
-        const wgpu::CommandEncoderDescriptor desc{ .label = "blit_deprecated_command" };
-        mCommandEncoder = mDevice.CreateCommandEncoder(&desc);
-    }
-
-    const WebGPUBlitter::BlitArgs blitArgs{
-        .source = { .texture = sourceTexture->getTexture(),
-                    .aspect = sourceTexture->getAspect(),
-                    .origin = sourceOrigin,
-                    .extent = sourceSize,
-                    .mipLevel = sourceAttachment.level,
-                    .layerOrDepth = sourceAttachment.layer,
-        },
-        .destination = { .texture = destinationTexture->getTexture(),
-                         .aspect = destinationTexture->getAspect(),
-                         .origin = destinationOrigin,
-                         .extent = destinationSize,
-                         .mipLevel = destinationAttachment.level,
-                         .layerOrDepth = destinationAttachment.layer,
-        },
-        .filter = filter,
-    };
-    mBlitter.blit(mQueue, mCommandEncoder, blitArgs);
-
-    if (!reusedCommandEncoder) {
-        const wgpu::CommandBufferDescriptor desc{ .label = "blit_deprecated_command_buffer" };
-        const wgpu::CommandBuffer blitCommand{ mCommandEncoder.Finish(&desc) };
-        mQueue.Submit(1, &blitCommand);
-        mCommandEncoder = nullptr;
-    }
+    PANIC_PRECONDITION("WebGPUDriver::blitDEPRECATED not supported");
 }
 
 void WebGPUDriver::resolve(Handle<HwTexture> destinationTextureHandle, const uint8_t sourceLevel,
         const uint8_t sourceLayer, Handle<HwTexture> sourceTextureHandle,
         const uint8_t destinationLevel, const uint8_t destinationLayer) {
-    FWGPU_SYSTRACE_SCOPE();
-    if (!mCommandEncoder) {
-        const wgpu::CommandEncoderDescriptor encoderDescriptor{
-            .label = "command_created_with_resolve",
-        };
-        mCommandEncoder = mDevice.CreateCommandEncoder(&encoderDescriptor);
-    }
+    FILAMENT_CHECK_PRECONDITION(mCommandEncoder)
+            << "Resolve assumes there is a valid command encoder to piggyback on.";
     FILAMENT_CHECK_PRECONDITION(mRenderPassEncoder == nullptr)
             << "Resolve cannot be called during an existing render pass";
     const auto sourceTexture{ handleCast<WebGPUTexture>(sourceTextureHandle) };
@@ -1747,7 +1511,6 @@ void WebGPUDriver::blit(Handle<HwTexture> destinationTextureHandle, const uint8_
         const uint8_t sourceLayer, const math::uint2 destinationOrigin,
         Handle<HwTexture> sourceTextureHandle, const uint8_t destinationLevel,
         const uint8_t destinationLayer, const math::uint2 sourceOrigin, const math::uint2 size) {
-    FWGPU_SYSTRACE_SCOPE();
     bool reusedCommandEncoder{ true };
     if (mCommandEncoder) {
         // make sure command elements (draws, etc.) prior to this blit are processed before the blit
@@ -1806,7 +1569,6 @@ void WebGPUDriver::blit(Handle<HwTexture> destinationTextureHandle, const uint8_
 }
 
 void WebGPUDriver::bindPipeline(PipelineState const& pipelineState) {
-    FWGPU_SYSTRACE_SCOPE();
     assert_invariant(mRenderPassEncoder);
     const auto program{ handleCast<WebGPUProgram>(pipelineState.program) };
     assert_invariant(program);
@@ -1843,10 +1605,13 @@ void WebGPUDriver::bindPipeline(PipelineState const& pipelineState) {
     };
     wgpu::TextureFormat depthStencilFormat{ wgpu::TextureFormat::Undefined };
     if (renderTarget->isDefaultRenderTarget()) {
+        // default render target color(s) (one)...
         colorFormatCount = 1;
         colorFormats[0] = mSwapChain->getColorFormat();
+        // default render target depth/stencil...
         depthStencilFormat = mSwapChain->getDepthFormat();
     } else {
+        // custom render target color(s)...
         MRT const& mrtColorAttachments{ mCurrentRenderTarget->getColorAttachmentInfos() };
         for (size_t i{ 0 }; i < MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT; ++i) {
             if (mrtColorAttachments[i].handle) {
@@ -1895,7 +1660,6 @@ void WebGPUDriver::bindPipeline(PipelineState const& pipelineState) {
 }
 
 void WebGPUDriver::bindRenderPrimitive(Handle<HwRenderPrimitive> renderPrimitiveHandle) {
-    FWGPU_SYSTRACE_SCOPE();
     const auto renderPrimitive = handleCast<WebGPURenderPrimitive>(renderPrimitiveHandle);
     const auto vertexBufferInfo = handleCast<WebGPUVertexBufferInfo>(
             renderPrimitive->vertexBuffer->getVertexBufferInfoHandle());
@@ -1913,8 +1677,7 @@ void WebGPUDriver::bindRenderPrimitive(Handle<HwRenderPrimitive> renderPrimitive
 
 void WebGPUDriver::draw2(const uint32_t indexOffset, const uint32_t indexCount,
         const uint32_t instanceCount) {
-    FWGPU_SYSTRACE_SCOPE();
-    // Bind groups are deferred until the actual draw call.
+    // We defer actually binding until we actually draw
     for (size_t i = 0; i < MAX_DESCRIPTOR_SET_COUNT; i++) {
         auto& binding = mCurrentDescriptorSets[i];
         if (binding.bindGroup) {
@@ -1960,7 +1723,6 @@ void WebGPUDriver::resetState(int) {
 void WebGPUDriver::updateDescriptorSetBuffer(Handle<HwDescriptorSet> descriptorSetHandle,
         const backend::descriptor_binding_t binding, Handle<HwBufferObject> bufferObjectHandle,
         const uint32_t offset, const uint32_t size) {
-    FWGPU_SYSTRACE_SCOPE();
     const auto bindGroup = handleCast<WebGPUDescriptorSet>(descriptorSetHandle);
     const auto buffer = handleCast<WebGPUBufferObject>(bufferObjectHandle);
     if (!bindGroup->getIsLocked()) {
@@ -1981,13 +1743,14 @@ void WebGPUDriver::updateDescriptorSetBuffer(Handle<HwDescriptorSet> descriptorS
 void WebGPUDriver::updateDescriptorSetTexture(Handle<HwDescriptorSet> descriptorSetHandle,
         const backend::descriptor_binding_t binding, Handle<HwTexture> textureHandle,
         const SamplerParams params) {
-    FWGPU_SYSTRACE_SCOPE();
     auto bindGroup = handleCast<WebGPUDescriptorSet>(descriptorSetHandle);
     auto texture = handleCast<WebGPUTexture>(textureHandle);
 
     if (!bindGroup->getIsLocked()) {
-        // Dawn will cache duplicate samplers, so we don't strictly need to maintain a cache here.
+        // Dawn will cache duplicate samplers, so we don't strictly need to maintain a cache.
+        //  Making a cache might save us minor perf by reducing param translation
         const auto sampler = makeSampler(params);
+        // TODO making assumptions that size and offset mean the same thing here.
         wgpu::BindGroupEntry tEntry{
             .binding = static_cast<uint32_t>(binding * 2),
             .textureView = texture->getDefaultTextureView() };
@@ -2021,7 +1784,7 @@ void WebGPUDriver::bindDescriptorSet(Handle<HwDescriptorSet> descriptorSetHandle
         const backend::descriptor_set_t setIndex, backend::DescriptorSetOffsetArray&& offsets) {
     assert_invariant(setIndex < MAX_DESCRIPTOR_SET_COUNT);
 
-    // An empty handle unbinds the descriptor set.
+    // An empty handle signifies we need to release this bind point
     if (!descriptorSetHandle) {
         mCurrentDescriptorSets[setIndex] = {};
         return;
@@ -2151,4 +1914,4 @@ wgpu::AddressMode WebGPUDriver::fWrapModeToWAddressMode(const SamplerWrapMode& f
     return wgpu::AddressMode::Undefined;
 }
 
-} // namespace filament::backend
+} // namespace filament

filament/backend/src/webgpu/WebGPUDriver.h

@@ -24,7 +24,6 @@
 #include "webgpu/WebGPUPipelineCache.h"
 #include "webgpu/WebGPUPipelineLayoutCache.h"
 #include "webgpu/WebGPURenderPassMipmapGenerator.h"
-#include "webgpu/utils/AsyncTaskCounter.h"
 #include <backend/platforms/WebGPUPlatform.h>
 
 #include "DriverBase.h"
@@ -48,12 +47,8 @@
 namespace filament::backend {
 
 class WebGPUSwapChain;
-
 /**
- * Implements the private backend driver API for WebGPU specifically (that API is essentially
- * expressed in private/backend/DriverAPI.inc)
- *
- * It manages all the WebGPU resources necessary to accomplish this.
+ * WebGPU backend (driver) implementation
  */
 class WebGPUDriver final : public DriverBase {
 public:
@@ -69,8 +64,8 @@ private:
     [[nodiscard]] wgpu::Sampler makeSampler(SamplerParams const& params);
     [[nodiscard]] static wgpu::AddressMode fWrapModeToWAddressMode(const filament::backend::SamplerWrapMode& fUsage);
 
-    // The platform (e.g. OS) specific aspects of the WebGPU backend are strictly only
-    // handled in the WebGPUPlatform.
+    // the platform (e.g. OS) specific aspects of the WebGPU backend are strictly only
+    // handled in the WebGPUPlatform
     WebGPUPlatform& mPlatform;
     wgpu::Adapter mAdapter = nullptr;
     wgpu::Device mDevice = nullptr;
@@ -91,7 +86,6 @@ private:
     spd::MipmapGenerator mSpdComputePassMipmapGenerator;
     WebGPUMsaaTextureResolver mMsaaTextureResolver{};
     WebGPUBlitter mBlitter;
-    webgpuutils::AsyncTaskCounter mReadPixelMapsCounter{};
 
     struct DescriptorSetBindingInfo{
         wgpu::BindGroup bindGroup;

filament/backend/src/webgpu/WebGPUFence.cpp

@@ -28,11 +28,12 @@ namespace filament::backend {
 FenceStatus WebGPUFence::getStatus() { return mStatus.load(); }
 
 void WebGPUFence::addMarkerToQueueState(wgpu::Queue const& queue) {
-    // The lambda function is called when the work is done. It updates the fence status based on the
-    // result of the work.
     queue.OnSubmittedWorkDone(
         wgpu::CallbackMode::AllowSpontaneous,
         [this](const wgpu::QueueWorkDoneStatus status, wgpu::StringView message) {
+            // Note: The 'message' parameter is required by the function signature
+            // but can be ignored if not needed.
+
             switch (status) {
                 case wgpu::QueueWorkDoneStatus::Success:
                     mStatus.store(FenceStatus::CONDITION_SATISFIED);

filament/backend/src/webgpu/WebGPUFence.h

@@ -28,11 +28,6 @@ class Queue;
 
 namespace filament::backend {
 
-/**
-  * A WebGPU-specific implementation of the HwFence.
-  * This class is used to synchronize the CPU and GPU. It allows the CPU to know when the GPU has
-  * finished a set of commands.
-  */
 class WebGPUFence final : public HwFence {
 public:
     [[nodiscard]] FenceStatus getStatus();
@@ -40,9 +35,6 @@ public:
     void addMarkerToQueueState(wgpu::Queue const&);
 
 private:
-    // The current status of the fence.
-    // This is atomic because it can be updated by a WebGPU callback thread and read from the main
-    // thread.
     std::atomic<FenceStatus> mStatus{ FenceStatus::TIMEOUT_EXPIRED };
 };
 

filament/backend/src/webgpu/WebGPUIndexBuffer.h

@@ -30,10 +30,6 @@ enum class IndexFormat : uint32_t;
 
 namespace filament::backend {
 
-/**
- * A WebGPU implementation of the HwIndexBuffer.
- * This class represents a GPU buffer that stores indices for indexed drawing.
- */
 class WebGPUIndexBuffer final : public HwIndexBuffer, public WebGPUBufferBase {
 public:
     WebGPUIndexBuffer(wgpu::Device const&, uint8_t elementSize, uint32_t indexCount);

filament/backend/src/webgpu/WebGPUMsaaTextureResolver.cpp

@@ -48,8 +48,8 @@ void resolveColorTextures(wgpu::CommandEncoder const& commandEncoder,
     const wgpu::RenderPassEncoder renderPassEncoder{ commandEncoder.BeginRenderPass(
             &renderPassDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(renderPassEncoder)
-            << "Failed to create wgpu::RenderPassEncoder for MSAA resolve.";
-    renderPassEncoder.End(); // The resolve is implicit in the pass.
+            << "Failed to create wgpu::RenderPassEncoder for WebGPUDriver::resolve";
+    renderPassEncoder.End(); // only the implicit resolve is happening in the pass
 }
 
 } // namespace
@@ -59,29 +59,30 @@ void WebGPUMsaaTextureResolver::resolve(ResolveRequest const& request) {
     ResolveRequest::TextureInfo const& destination{ request.destination };
     FILAMENT_CHECK_PRECONDITION(destination.texture.GetWidth() == source.texture.GetWidth() &&
                                 destination.texture.GetHeight() == source.texture.GetHeight())
-            << "Invalid resolve: source and destination sizes don't match.";
+            << "invalid resolve: source and destination sizes don't match";
 
     FILAMENT_CHECK_PRECONDITION(
             source.texture.GetSampleCount() > 1 && destination.texture.GetSampleCount() == 1)
-            << "Invalid resolve: source.samples=" << source.texture.GetSampleCount()
+            << "invalid resolve: source.samples=" << source.texture.GetSampleCount()
             << ", destination.samples=" << destination.texture.GetSampleCount();
 
     FILAMENT_CHECK_PRECONDITION(source.texture.GetFormat() == destination.texture.GetFormat())
-            << "Source and destination texture format don't match.";
+            << "source and destination texture format don't match";
     const wgpu::TextureFormat format{ source.texture.GetFormat() };
 
-    FILAMENT_CHECK_PRECONDITION(!hasDepth(format)) << "Can't resolve depth formats.";
-    FILAMENT_CHECK_PRECONDITION(!hasStencil(format)) << "Can't resolve stencil formats.";
+    FILAMENT_CHECK_PRECONDITION(!hasDepth(format)) << "can't resolve depth formats";
+
+    FILAMENT_CHECK_PRECONDITION(!hasStencil(format)) << "can't resolve stencil formats";
 
     FILAMENT_CHECK_PRECONDITION(source.texture.GetUsage() & wgpu::TextureUsage::RenderAttachment)
-            << "Source texture usage doesn't have wgpu::TextureUsage::RenderAttachment.";
+            << "source texture usage doesn't have wgpu::TextureUsage::RenderAttachment";
 
     FILAMENT_CHECK_PRECONDITION(
             destination.texture.GetUsage() & wgpu::TextureUsage::RenderAttachment)
-            << "Destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment.";
+            << "destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment";
 
     FILAMENT_CHECK_PRECONDITION(destination.texture.GetUsage() & wgpu::TextureUsage::TextureBinding)
-            << "Destination texture usage doesn't have wgpu::TextureUsage::TextureBinding.";
+            << "destination texture usage doesn't have wgpu::TextureUsage::TextureBinding";
 
     const wgpu::TextureViewDescriptor sourceTextureViewDescriptor{
         .label = "resolve_source_texture_view",
@@ -115,7 +116,7 @@ void WebGPUMsaaTextureResolver::resolve(ResolveRequest const& request) {
             << "Failed to create wgpu::TextureView destinationTextureView.";
 
     if (hasDepth(format)) {
-        PANIC_PRECONDITION("Depth resolve not yet implemented.");
+        PANIC_PRECONDITION("DEPTH RESOLVE NOT IMPLEMENTED YET");
     } else {
         resolveColorTextures(request.commandEncoder, sourceTextureView, destinationTextureView);
     }

filament/backend/src/webgpu/WebGPUMsaaTextureResolver.h

@@ -31,10 +31,6 @@ namespace filament::backend {
 
 class WebGPUTexture;
 
-/**
- * A utility class for resolving multisampled textures in WebGPU.
- * This is necessary because WebGPU does not have an implicit resolve step like other APIs.
- */
 class WebGPUMsaaTextureResolver final {
 public:
     struct ResolveRequest final {
@@ -51,9 +47,6 @@ public:
         TextureInfo destination;
     };
 
-    /**
-     * Resolves a multisampled texture into a single-sampled texture.
-     */
     void resolve(ResolveRequest const&);
 };
 

filament/backend/src/webgpu/WebGPUPipelineCache.cpp

@@ -54,7 +54,7 @@ namespace {
         case CullingMode::FRONT: return wgpu::CullMode::Front;
         case CullingMode::BACK:  return wgpu::CullMode::Back;
         case CullingMode::FRONT_AND_BACK:
-            // WebGPU does not support culling both front and back faces simultaneously.
+            // no WegGPU equivalent of front and back
             FILAMENT_CHECK_POSTCONDITION(false)
                     << "WebGPU does not support CullingMode::FRONT_AND_BACK";
             return wgpu::CullMode::Undefined;
@@ -220,7 +220,7 @@ wgpu::RenderPipeline WebGPUPipelineCache::createRenderPipeline(
     const bool requestedDepth{ any(request.targetRenderFlags & TargetBufferFlags::DEPTH) };
     const bool requestedStencil{ any(request.targetRenderFlags & TargetBufferFlags::STENCIL) };
     const bool depthOrStencilRequested{ requestedDepth || requestedStencil };
-
+    // depth/stencil...
     if (depthOrStencilRequested) {
         FILAMENT_CHECK_PRECONDITION(request.depthStencilFormat != wgpu::TextureFormat::Undefined)
                 << "Depth or Stencil requested for pipeline, but depthStencilFormat is "
@@ -293,11 +293,10 @@ wgpu::RenderPipeline WebGPUPipelineCache::createRenderPipeline(
         },
         .primitive = {
             .topology = toWebGPU(request.primitiveType),
-            // TODO This won't always be the case for the primitives bound.  But this particular field
-            //  in the pipeline struct is only meant for restarting the strip, which we don't need so far.
-            .stripIndexFormat =
-                    isStripPrimitiveType(request.primitiveType)? wgpu::IndexFormat::Uint16 :
-                                                                wgpu::IndexFormat::Undefined,
+            // TODO should we assume some constant format here or is there a way to get
+            //      this from PipelineState somehow or elsewhere?
+            //      Perhaps, cache/assert format from index buffers as they are requested?
+            .stripIndexFormat = wgpu::IndexFormat::Undefined,
             .frontFace = request.rasterState.inverseFrontFaces ? wgpu::FrontFace::CW : wgpu::FrontFace::CCW,
             .cullMode = toWebGPU(request.rasterState.culling),
             // TODO no depth clamp in WebGPU supported directly. unclippedDepth is close, so we are
@@ -313,11 +312,6 @@ wgpu::RenderPipeline WebGPUPipelineCache::createRenderPipeline(
         },
         .fragment = nullptr // will add below if fragment module is included
     };
-    FILAMENT_CHECK_POSTCONDITION(!isStripPrimitiveType(request.primitiveType) ||
-            pipelineDescriptor.primitive.stripIndexFormat != wgpu::IndexFormat::Undefined)
-            << "If the topology is a strip format, e.g. TriangleStrip, the stripIndexFormat cannot "
-               "be Undefined.";
-
     wgpu::FragmentState fragmentState = {};
     const wgpu::BlendState blendState {
         .color = {
@@ -371,7 +365,6 @@ bool WebGPUPipelineCache::RenderPipelineKeyEqual::operator()(RenderPipelineKey c
 }
 
 void WebGPUPipelineCache::removeExpiredPipelines() {
-    FWGPU_SYSTRACE_SCOPE();
     using Iterator = decltype(mRenderPipelines)::const_iterator;
     for (Iterator iterator{ mRenderPipelines.begin() }; iterator != mRenderPipelines.end();) {
         RenderPipelineCacheEntry const& entry{ iterator.value() };

filament/backend/src/webgpu/WebGPUPipelineCache.h

@@ -34,11 +34,6 @@
 
 namespace filament::backend {
 
-/**
- * A cache for WebGPU render pipelines.
- * This class is responsible for creating and caching wgpu::RenderPipeline objects to avoid
- * expensive pipeline creation at runtime.
- */
 class WebGPUPipelineCache final {
 public:
     struct RenderPipelineRequest final {
@@ -68,9 +63,6 @@ public:
     [[nodiscard]] wgpu::RenderPipeline const& getOrCreateRenderPipeline(
             RenderPipelineRequest const&);
 
-    /**
-     * Should be called at the end of each frame to perform cache maintenance.
-     */
     void onFrameEnd();
 
 private:
@@ -161,7 +153,7 @@ private:
         // render targets...                                                                //
         uint8_t multisampleCount{ 0 };                                                      // 1    : 313
         uint8_t colorFormatCount{ 0 };                                                      // 1    : 314
-        uint8_t padding[5]{ 0 };                                                            // 5    : 315
+        uint8_t padding[5]{ 0 };                                                            // 5    : 319
         TargetBufferFlags targetRenderFlags{ TargetBufferFlags::NONE };                     // 4    : 320
         wgpu::TextureFormat depthStencilFormat { wgpu::TextureFormat::Undefined };          // 4    : 324
         wgpu::TextureFormat colorFormats[MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT]{           //

filament/backend/src/webgpu/WebGPUPipelineLayoutCache.cpp

@@ -77,27 +77,26 @@ wgpu::PipelineLayout WebGPUPipelineLayoutCache::createPipelineLayout(
     };
     const wgpu::PipelineLayout layout{ mDevice.CreatePipelineLayout(&descriptor) };
     FILAMENT_CHECK_POSTCONDITION(layout)
-            << "Failed to create pipeline layout " << descriptor.label << ".";
+            << "Failed to create pipeline layout " << descriptor.label << "?";
     return layout;
 }
 
 bool WebGPUPipelineLayoutCache::PipelineLayoutKeyEqual::operator()(PipelineLayoutKey const& key1,
         PipelineLayoutKey const& key2) const {
-    // Compare the raw bytes of the keys for equality.
     return 0 == memcmp(reinterpret_cast<void const*>(&key1), reinterpret_cast<void const*>(&key2),
                         sizeof(key1));
 }
 
 void WebGPUPipelineLayoutCache::removeExpiredPipelineLayouts() {
-    FWGPU_SYSTRACE_SCOPE();
     using Iterator = decltype(mPipelineLayouts)::const_iterator;
     for (Iterator iterator{ mPipelineLayouts.begin() }; iterator != mPipelineLayouts.end();) {
         PipelineLayoutCacheEntry const& entry{ iterator.value() };
         if (mFrameCount > (entry.lastUsedFrameCount +
                                   FILAMENT_WEBGPU_PIPELINE_LAYOUT_EXPIRATION_IN_FRAME_COUNT)) {
-            // The pipeline layout has not been used recently, so we can remove it from the cache.
+            // pipeline layout expired...
             iterator = mPipelineLayouts.erase(iterator);
         } else {
+            // pipeline layout not yet expired...
             ++iterator;
         }
     }

filament/backend/src/webgpu/WebGPUPipelineLayoutCache.h

@@ -31,11 +31,6 @@
 
 namespace filament::backend {
 
-/**
- * A cache for WebGPU pipeline layouts.
- * This class is responsible for creating and caching wgpu::PipelineLayout objects to avoid
- * expensive pipeline layout creation at runtime.
- */
 class WebGPUPipelineLayoutCache final {
 public:
     struct PipelineLayoutRequest final {
@@ -50,16 +45,9 @@ public:
     WebGPUPipelineLayoutCache& operator=(WebGPUPipelineLayoutCache const&) = delete;
     WebGPUPipelineLayoutCache& operator=(WebGPUPipelineLayoutCache const&&) = delete;
 
-    /**
-     * Retrieves a pipeline layout from the cache or creates a new one if it doesn't exist.
-     * @return A constant reference to the cached or newly created pipeline layout.
-     */
     [[nodiscard]] wgpu::PipelineLayout const& getOrCreatePipelineLayout(
             PipelineLayoutRequest const&);
 
-    /**
-     * Should be called at the end of each frame to perform cache maintenance.
-     */
     void onFrameEnd();
 
 private:
@@ -71,10 +59,10 @@ private:
      *  instances, single bytes for booleans etc.), trivial copying and comparison (byte by byte),
      *  and a word-aligned structure with a size in bytes as a multiple of 4 (for murmer hash).
      */
-    struct PipelineLayoutKey final {
-        WGPUBindGroupLayout bindGroupLayoutHandles[MAX_DESCRIPTOR_SET_COUNT]{ nullptr }; // 32 :0
-        uint8_t bindGroupLayoutCount{ 0 };                                               // 1  :32
-        uint8_t padding[7]{ 0 };                                                         // 7  :33
+    struct PipelineLayoutKey final {                                                     // size : offset (need multiples of 4 bytes for hashing)
+        WGPUBindGroupLayout bindGroupLayoutHandles[MAX_DESCRIPTOR_SET_COUNT]{ nullptr }; // 32   : 0
+        uint8_t bindGroupLayoutCount{ 0 };                                               // 1    : 32
+        uint8_t padding[7]{ 0 };                                                         // 7    : 33
     };
     static_assert(sizeof(PipelineLayoutKey) == 40,
             "PipelineLayoutKey must not have implicit padding.");

filament/backend/src/webgpu/WebGPUProgram.cpp

@@ -89,7 +89,7 @@ namespace {
                 std::string_view(sourceData + posAfterId, posEndOfStatement - posAfterId);
         size_t posOfEqual = statementSegment.find('=');
         if (posOfEqual == std::string::npos) {
-            // Not an assignment, so we don't need to replace it.
+            // not an assignment statement, so stream to the end of the statement and continue...
             processedShaderSource << std::string_view(sourceData + pos,
                     posEndOfStatement + 1 - pos);
             pos = posEndOfStatement + 1;
@@ -122,7 +122,10 @@ namespace {
         }
         const auto newValueItr = specConstants.find(static_cast<uint32_t>(constantId));
         if (newValueItr == specConstants.end()) {
-            // The constant is not being overridden, so keep the default value.
+            // not going to override the constant,
+            // as the specConstants parameter doesn't specify it. So, we will keep the default
+            // already in the source text
+            // (stream to the end of the statement)...
             processedShaderSource << std::string_view(sourceData + pos,
                     posEndOfStatement + 1 - pos);
             pos = posEndOfStatement + 1;
@@ -171,7 +174,7 @@ namespace {
     utils::FixedCapacityVector<uint8_t> const& sourceBytes =
             shaderSource[static_cast<size_t>(stage)];
     if (sourceBytes.empty()) {
-        return nullptr; // No shader source to compile.
+        return nullptr;// nothing to compile/create, the shader was not provided
     }
     std::stringstream labelStream;
     labelStream << programName << " " << filamentShaderStageToString(stage) << " shader";
@@ -189,8 +192,7 @@ namespace {
     };
     const wgpu::ShaderModule shaderModule = device.CreateShaderModule(&descriptor);
     const wgpu::Instance instance = device.GetAdapter().GetInstance();
-
-    // Synchronously compile the shader module.
+    // synchronously creates the shader module...
     const wgpu::WaitStatus waitResult = instance.WaitAny(
             shaderModule.GetCompilationInfo(wgpu::CallbackMode::WaitAnyOnly,
                     [&descriptor](auto const& status,
@@ -280,7 +282,7 @@ WebGPUProgram::WebGPUProgram(wgpu::Device const& device, Program const& program)
     : HwProgram{ program.getName() } {
     std::unordered_map<uint32_t, std::variant<int32_t, float, bool>> specConstants;
     toMap(program.getSpecializationConstants(), specConstants);
-    // TODO: Consider creating/compiling these shaders in parallel.
+    // TODO consider creating/compiling these shaders in parallel
     vertexShaderModule = createShaderModule(device, program, ShaderStage::VERTEX, specConstants);
     fragmentShaderModule =
             createShaderModule(device, program, ShaderStage::FRAGMENT, specConstants);

filament/backend/src/webgpu/WebGPUProgram.h

@@ -25,10 +25,6 @@ namespace filament::backend {
 
 class Program;
 
-/**
- * A WebGPU implementation of the HwProgram.
- * This class holds the compiled shader modules for a given program.
- */
 class WebGPUProgram final : public HwProgram {
 public:
     WebGPUProgram(wgpu::Device const&, Program const&);

filament/backend/src/webgpu/WebGPURenderPassMipmapGenerator.cpp

@@ -27,8 +27,9 @@ namespace filament::backend {
 
 namespace {
 
-// The shaders expect a single-sampled 2D texture with a format compatible with a scalar sample format of f32.
-// This is checked with the getCompatibilityFor function.
+// The shaders expect a single-sampled 2D texture
+// with a format compatible with scalar sample format f32
+// (this is checked with the getCompatibilityFor function)
 constexpr uint32_t TEXTURE_BIND_GROUP_INDEX{ 0 };
 constexpr size_t TEXTURE_BIND_GROUP_ENTRY_SIZE{ 2 }; // sampler and texture
 constexpr uint32_t SAMPLER_BINDING_INDEX{ 0 };
@@ -84,7 +85,7 @@ constexpr std::string_view SHADER_SOURCE{ R"(
         .maxAnisotropy = 1, // should not matter, just being consistently defined
     };
     const wgpu::Sampler sampler{ device.CreateSampler(&descriptor) };
-    FILAMENT_CHECK_POSTCONDITION(sampler) << "Failed to create previous mip level sampler.";
+    FILAMENT_CHECK_POSTCONDITION(sampler) << "Failed to create previous mip level sampler?";
     return sampler;
 }
 
@@ -97,7 +98,7 @@ constexpr std::string_view SHADER_SOURCE{ R"(
     };
     const wgpu::ShaderModule shaderModule{ device.CreateShaderModule(&shaderModuleDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(shaderModule)
-            << "Failed to create shader module for render pass mipmap generation.";
+            << "Failed to create shader module for render pass mipmap generation?";
     return shaderModule;
 }
 
@@ -130,7 +131,7 @@ constexpr std::string_view SHADER_SOURCE{ R"(
     const wgpu::BindGroupLayout textureBindGroupLayout{ device.CreateBindGroupLayout(
             &textureBindGroupLayoutDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(textureBindGroupLayout)
-            << "Failed to create texture bind group layout for render pass mipmap generation.";
+            << "Failed to create texture bind group layout for render pass mipmap generation?";
     return textureBindGroupLayout;
 }
 
@@ -144,7 +145,7 @@ constexpr std::string_view SHADER_SOURCE{ R"(
     const wgpu::PipelineLayout pipelineLayout{ device.CreatePipelineLayout(
             &pipelineLayoutDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(pipelineLayout)
-            << "Failed to create pipeline layout for render pass mipmap generation.";
+            << "Failed to create pipeline layout for render pass mipmap generation?";
     return pipelineLayout;
 }
 
@@ -193,7 +194,7 @@ constexpr std::string_view SHADER_SOURCE{ R"(
     };
     const wgpu::RenderPipeline pipeline{ device.CreateRenderPipeline(&pipelineDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(pipeline)
-            << "Failed to create pipeline for render pass mipmap generation.";
+            << "Failed to create pipeline for render pass mipmap generation?";
     return pipeline;
 }
 
@@ -209,7 +210,8 @@ WebGPURenderPassMipmapGenerator::WebGPURenderPassMipmapGenerator(wgpu::Device co
 WebGPURenderPassMipmapGenerator::FormatCompatibility
 WebGPURenderPassMipmapGenerator::getCompatibilityFor(const wgpu::TextureFormat format,
         const wgpu::TextureDimension dimension, const uint32_t sampleCount) {
-    // Check that the format is compatible (currently expects that the scalar sample type is f32).
+    // check that the format is compatible (currently expects that the scalar sample type is
+    // f32)...
     switch (format) {
         case wgpu::TextureFormat::Depth16Unorm:
         case wgpu::TextureFormat::Depth24Plus:
@@ -218,16 +220,20 @@ WebGPURenderPassMipmapGenerator::getCompatibilityFor(const wgpu::TextureFormat f
         case wgpu::TextureFormat::Depth32FloatStencil8:
             return {
                 .compatible = false,
-                .reason = "A depth texture format requires special sampler treatment and does not "
-                          "generally support linear filtering needed for render pass based mipmap "
-                          "generation.",
+                .reason = "A depth texture format requires special sampler treatment and does "
+                          "not "
+                          "generally support linear filtering needed for render pass based "
+                          "mipmap "
+                          "generation, thus this texture is not supported, as it is a kind of "
+                          "depth texture.",
             };
         case wgpu::TextureFormat::Undefined:
         case wgpu::TextureFormat::External:
             return {
                 .compatible = false,
-                .reason = "Undefined or External textures are not supported for render pass based "
-                          "mipmap generation.",
+                .reason = "Undefined or External textures are not supported for render pass "
+                          "based "
+                          "mipmap generation",
             };
         default:
             const ScalarSampleType scalarSampleType{ getScalarSampleTypeFrom(format) };
@@ -236,19 +242,21 @@ WebGPURenderPassMipmapGenerator::getCompatibilityFor(const wgpu::TextureFormat f
                     return {
                         .compatible = false,
                         .reason = "The provided texture format requires an unsigned integer "
-                                  "sampler, which does not natively support linear filtering.",
+                                  "sampler, which does not natively support linear filtering, "
+                                  "which is needed for render pass based mipmap generation.",
                     };
                 case ScalarSampleType::I32:
                     return {
                         .compatible = false,
                         .reason = "The provided texture format requires a signed integer "
-                                  "sampler, which does not natively support linear filtering.",
+                                  "sampler, which does not natively support linear filtering, "
+                                  "which is needed for render pass based mipmap generation.",
                     };
                 case ScalarSampleType::F32:
                     break;
             }
     }
-    // Check that the dimensionality is compatible (currently expects 2D textures).
+    // check that the dimensionality is compatible (currently expects 2D textures)...
     switch (dimension) {
         case wgpu::TextureDimension::Undefined:
             return {
@@ -274,7 +282,7 @@ WebGPURenderPassMipmapGenerator::getCompatibilityFor(const wgpu::TextureFormat f
         case wgpu::TextureDimension::e2D:
             break;
     }
-    // Check that the texture is single-sampled.
+    // check that the texture is single-sampled (for now)...
     if (sampleCount > 1) {
         return {
             .compatible = false,
@@ -419,7 +427,7 @@ void WebGPURenderPassMipmapGenerator::generateMipmaps(wgpu::Queue const& queue,
         wgpu::Texture const& texture) {
     const uint32_t mipLevelCount{ texture.GetMipLevelCount() };
     if (mipLevelCount < 2) {
-        return; // Nothing to do.
+        return; // nothing to do
     }
     wgpu::RenderPipeline const& pipeline{ getOrCreatePipelineFor(texture.GetFormat()) };
     const wgpu::CommandEncoderDescriptor commandEncoderDescriptor{
@@ -428,19 +436,20 @@ void WebGPURenderPassMipmapGenerator::generateMipmaps(wgpu::Queue const& queue,
     const wgpu::CommandEncoder commandEncoder{ mDevice.CreateCommandEncoder(
             &commandEncoderDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(commandEncoder)
-            << "Failed to create command encoder for render pass mipmap generation.";
+            << "Failed to create command encoder for layer  for render pass mipmap generation?";
     const uint32_t layerCount{ texture.GetDepthOrArrayLayers() };
     for (uint32_t layer = 0; layer < layerCount; layer++) {
         for (uint32_t mipLevel = 1; mipLevel < mipLevelCount; mipLevel++) {
             generateMipmap(commandEncoder, texture, pipeline, layer, mipLevel);
         }
     }
+    // submit the command buffer...
     const wgpu::CommandBufferDescriptor commandBufferDescriptor{
         .label = "mipmap_generation_render_pass_cmd_buffer",
     };
     const wgpu::CommandBuffer commandBuffer{ commandEncoder.Finish(&commandBufferDescriptor) };
     FILAMENT_CHECK_POSTCONDITION(commandBuffer)
-            << "Failed to create command buffer for render pass mipmap generation.";
+            << "Failed to create command buffer for layer  for render pass mipmap generation?";
     queue.Submit(1, &commandBuffer);
 }
 
@@ -463,7 +472,7 @@ void WebGPURenderPassMipmapGenerator::generateMipmap(wgpu::CommandEncoder const&
         .usage = wgpu::TextureUsage::TextureBinding,
     };
     const wgpu::TextureView sourceView{ texture.CreateView(&sourceViewDescriptor) };
-    FILAMENT_CHECK_POSTCONDITION(sourceView)
+    FILAMENT_CHECK_POSTCONDITION(pipeline)
             << "Failed to create source texture view for layer " << layer << " and mip level "
             << mipLevel << " for render pass mipmap generation?";
     const wgpu::TextureViewDescriptor destinationViewDescriptor{
@@ -478,7 +487,7 @@ void WebGPURenderPassMipmapGenerator::generateMipmap(wgpu::CommandEncoder const&
         .usage = wgpu::TextureUsage::RenderAttachment,
     };
     const wgpu::TextureView destinationView{ texture.CreateView(&destinationViewDescriptor) };
-    FILAMENT_CHECK_POSTCONDITION(destinationView)
+    FILAMENT_CHECK_POSTCONDITION(pipeline)
             << "Failed to create destination texture view for layer " << layer << " and mip level "
             << mipLevel << " for render pass mipmap generation?";
     // create the render pass...
@@ -499,7 +508,7 @@ void WebGPURenderPassMipmapGenerator::generateMipmap(wgpu::CommandEncoder const&
         .entries = textureBindGroupEntries,
     };
     const wgpu::BindGroup textureBindGroup{ mDevice.CreateBindGroup(&textureBindGroupDescriptor) };
-    FILAMENT_CHECK_POSTCONDITION(textureBindGroup)
+    FILAMENT_CHECK_POSTCONDITION(pipeline)
             << "Failed to create texture bind group for layer " << layer << " and mip level "
             << mipLevel << " for render pass mipmap generation?";
     const wgpu::RenderPassColorAttachment colorAttachment{
@@ -525,7 +534,8 @@ void WebGPURenderPassMipmapGenerator::generateMipmap(wgpu::CommandEncoder const&
             << mipLevel << " for render pass mipmap generation?";
     renderPassEncoder.SetPipeline(pipeline);
     renderPassEncoder.SetBindGroup(TEXTURE_BIND_GROUP_INDEX, textureBindGroup);
-    renderPassEncoder.Draw(3); // Draw a full-screen triangle.
+    renderPassEncoder.Draw(3); // draw the full-screen triangle
+                                          // with hard-coded vertices in the shader
     renderPassEncoder.End();
 }
 

filament/backend/src/webgpu/WebGPURenderPassMipmapGenerator.h

@@ -27,9 +27,6 @@
 
 namespace filament::backend {
 
-/**
- * A utility class for generating mipmaps for a texture using a series of render passes.
- */
 class WebGPURenderPassMipmapGenerator final {
 public:
     struct FormatCompatibility final {
@@ -37,11 +34,6 @@ public:
         std::string_view reason;
     };
 
-    /**
-     * Checks if a given texture format is compatible with render pass-based mipmap generation.
-     * @return A FormatCompatibility struct indicating whether the format is compatible
-     *         and a reason if not.
-     */
     [[nodiscard]] static FormatCompatibility getCompatibilityFor(wgpu::TextureFormat,
             wgpu::TextureDimension, uint32_t sampleCount);
 

filament/backend/src/webgpu/WebGPURenderPrimitive.h

@@ -24,10 +24,6 @@ namespace filament::backend {
 class WebGPUVertexBuffer;
 class WebGPUIndexBuffer;
 
-/**
- * A WebGPU implementation of the HwRenderPrimitive.
- * This struct holds the vertex and index buffers that define a renderable primitive.
- */
 struct WebGPURenderPrimitive : public HwRenderPrimitive {
     WebGPUVertexBuffer* vertexBuffer = nullptr;
     WebGPUIndexBuffer* indexBuffer = nullptr;

filament/backend/src/webgpu/WebGPURenderTarget.cpp

@@ -41,9 +41,9 @@ namespace filament::backend {
 namespace {
 
 /**
- * Creates MSAA sidecar textures for attachments if necessary.
- * This function also verifies that all attachments have the same sample count.
- * @return The number of samples per attachment.
+ * Panics if the number of samples in the original attachment textures do not match.
+ * @return The number of samples in the original attachment textures (the number/count in each one).
+ *         Returns 0 if there are no attachments.
  */
 [[nodiscard]] uint8_t createMsaaSidecarTextures(const uint8_t renderTargetSampleCount,
         const TargetBufferFlags targetFlags, MRT const& colorAttachments,
@@ -100,6 +100,9 @@ namespace {
                 sampleCountPerAttachment = texture->samples;
                 firstAttachment = false;
             }
+            FILAMENT_CHECK_PRECONDITION(texture->samples == sampleCountPerAttachment)
+                    << target.name << " attachment texture has " << +texture->samples
+                    << " but the other attachment(s) have " << +sampleCountPerAttachment;
             if (renderTargetSampleCount > 1 && sampleCountPerAttachment == 1) {
                 texture->createMsaaSidecarTextureIfNotAlreadyCreated(renderTargetSampleCount,
                         device);
@@ -133,7 +136,7 @@ WebGPURenderTarget::WebGPURenderTarget(const uint32_t width, const uint32_t heig
     mColorAttachmentDesc.reserve(MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT);
 }
 
-// Constructor for the default render target
+// Default constructor for the default render target
 WebGPURenderTarget::WebGPURenderTarget()
     : HwRenderTarget{ 0, 0 },
       mDefaultRenderTarget{ true },
@@ -179,19 +182,19 @@ void WebGPURenderTarget::setUpRenderPassAttachments(wgpu::RenderPassDescriptor&
         FILAMENT_CHECK_PRECONDITION(std::all_of(customColorMsaaSidecarTextureViews,
                 customColorMsaaSidecarTextureViews + customColorTextureViewCount,
                 [](wgpu::TextureView const& msaaView) { return msaaView != nullptr; }))
-                << "A color or depth/stencil attachment texture has an MSAA sidecar but at least "
+                << "A color or depth/stencil attachment texture has a MSAA sidecar but at least "
                    "one other color attachment texture does not.";
         FILAMENT_CHECK_PRECONDITION(customDepthStencilMsaaSidecarTextureView != nullptr)
-                << "The color attachment texture(s) have MSAA sidecars but the depth/stencil "
+                << "The color attachment texture(s) have MSAA sidecar(s) but the depth/stencil "
                    "texture does not.";
     } else {
         FILAMENT_CHECK_PRECONDITION(std::all_of(customColorMsaaSidecarTextureViews,
                 customColorMsaaSidecarTextureViews + customColorTextureViewCount,
                 [](wgpu::TextureView const& msaaView) { return msaaView == nullptr; }))
-                << "A color or depth/stencil attachment texture does not have an MSAA sidecar but "
+                << "A color or depth/stencil attachment texture does not have a MSAA sidecar but "
                    "at least one color attachment texture does.";
         FILAMENT_CHECK_PRECONDITION(customDepthStencilMsaaSidecarTextureView == nullptr)
-                << "Custom color textures for the render target do not have MSAA sidecars but "
+                << "Custom color textures for the render target do not have MSAA sidecar(s) but "
                    "the depth/stencil texture does.";
     }
 

filament/backend/src/webgpu/WebGPURenderTarget.h

@@ -31,10 +31,6 @@
 
 namespace filament::backend {
 
-/**
- * A WebGPU implementation of the HwRenderTarget.
- * This class represents a collection of attachments (textures) that can be rendered to.
- */
 class WebGPURenderTarget : public HwRenderTarget {
 public:
     using Attachment = TargetBufferInfo; // Using TargetBufferInfo directly for attachments
@@ -42,8 +38,7 @@ public:
     WebGPURenderTarget(uint32_t width, uint32_t height, uint8_t samples, uint8_t layerCount,
             MRT const& colorAttachments, Attachment const& depthAttachment,
             Attachment const& stencilAttachment, TargetBufferFlags const& targetFlags,
-            std::function<WebGPUTexture*(const Handle<HwTexture>)> const&,
-            wgpu::Device const&);
+            std::function<WebGPUTexture*(const Handle<HwTexture>)> const&, wgpu::Device const&);
 
     // Default constructor for the default render target
     WebGPURenderTarget();
@@ -85,7 +80,7 @@ private:
 
     MRT mColorAttachments{};
     // TODO WebGPU only supports a DepthStencil attachment, should this be just
-    // mDepthStencilAttachment?
+    //      mDepthStencilAttachment?
     Attachment mDepthAttachment{};
     Attachment mStencilAttachment{};
 

filament/backend/src/webgpu/WebGPUSwapChain.cpp

@@ -187,8 +187,7 @@ void initConfig(wgpu::SurfaceConfiguration& config, wgpu::Device const& device,
         wgpu::SurfaceCapabilities const& capabilities, wgpu::Extent2D const& extent,
         bool useSRGBColorSpace) {
     config.device = device;
-    config.usage = wgpu::TextureUsage::RenderAttachment | wgpu::TextureUsage::CopySrc |
-                   wgpu::TextureUsage::TextureBinding;
+    config.usage = wgpu::TextureUsage::RenderAttachment | wgpu::TextureUsage::CopySrc;
     config.width = extent.width;
     config.height = extent.height;
     config.format = selectColorFormat(capabilities.formatCount, capabilities.formats, useSRGBColorSpace);
@@ -270,6 +269,8 @@ WebGPUSwapChain::WebGPUSwapChain(wgpu::Surface&& surface, wgpu::Extent2D const&
     const bool useSRGBColorSpace = (flags & SWAP_CHAIN_CONFIG_SRGB_COLORSPACE) != 0;
 
     initConfig(mConfig, device, capabilities, extent, useSRGBColorSpace);
+    mDepthFormat = selectDepthFormat(device.HasFeature(wgpu::FeatureName::Depth32FloatStencil8),
+            mNeedStencil);
 #if FWGPU_ENABLED(FWGPU_PRINT_SYSTEM)
     printSurfaceConfiguration(mConfig, mDepthFormat);
 #endif
@@ -296,6 +297,8 @@ WebGPUSwapChain::WebGPUSwapChain(wgpu::Extent2D const& extent,
     mConfig.presentMode = wgpu::PresentMode::Fifo;
     mConfig.alphaMode = wgpu::CompositeAlphaMode::Auto;
 
+    mDepthFormat = selectDepthFormat(device.HasFeature(wgpu::FeatureName::Depth32FloatStencil8),
+                mNeedStencil);
     const wgpu::TextureDescriptor textureDescriptor = {
         .usage = wgpu::TextureUsage::RenderAttachment | wgpu::TextureUsage::CopySrc,
         .dimension = wgpu::TextureDimension::e2D,

filament/backend/src/webgpu/WebGPUTexture.cpp

@@ -107,6 +107,84 @@ namespace {
     }
 }
 
+/**
+ * @param fUsage Filament's requested texture usage
+ * @param samples How many samples to use for MSAA
+ * @param needsComputeStorageSupport if we need to use this texture as storage binding in something
+ *                                   like a compute shader
+ * @param needsRenderAttachmentSupport if we need to use this texture as a render pass attachment
+ *                                     in something like a render pass blit (e.g. mipmap generation)
+ * @return The appropriate texture usage flags for the underlying texture
+ */
+[[nodiscard]] wgpu::TextureUsage fToWGPUTextureUsage(TextureUsage const& fUsage,
+        const uint8_t samples, const bool needsComputeStorageSupport,
+        const bool needsRenderAttachmentSupport) {
+    wgpu::TextureUsage retUsage = wgpu::TextureUsage::None;
+
+    // if needsComputeStorageSupport we need to read and write to the texture in a shader and thus
+    // require CopySrc, CopyDst, TextureBinding, & StorageBinding
+
+    // Basing this mapping off of VulkanTexture.cpp's getUsage func and suggestions from Gemini
+    // TODO Validate assumptions, revisit if issues.
+    if (needsComputeStorageSupport || any(TextureUsage::BLIT_SRC & fUsage)) {
+        retUsage |= wgpu::TextureUsage::CopySrc;
+    }
+    if (needsComputeStorageSupport ||
+            any((TextureUsage::BLIT_DST | TextureUsage::UPLOADABLE) & fUsage)) {
+        retUsage |= wgpu::TextureUsage::CopyDst;
+    }
+    if (needsComputeStorageSupport || any(TextureUsage::SAMPLEABLE & fUsage)) {
+        retUsage |= wgpu::TextureUsage::TextureBinding;
+    }
+    if (needsComputeStorageSupport) {
+        retUsage |= wgpu::TextureUsage::StorageBinding;
+    }
+    if (needsRenderAttachmentSupport) {
+        retUsage |= wgpu::TextureUsage::RenderAttachment;
+    }
+    if (any(TextureUsage::BLIT_SRC & fUsage)) {
+        retUsage |= wgpu::TextureUsage::TextureBinding;
+    }
+    if (any(TextureUsage::BLIT_DST & fUsage)) {
+        retUsage |= wgpu::TextureUsage::RenderAttachment;
+    }
+    if (any(TextureUsage::GEN_MIPMAPPABLE & fUsage)) {
+        retUsage |= (wgpu::TextureUsage::RenderAttachment | wgpu::TextureUsage::TextureBinding);
+    }
+    // WGPU Render attachment covers either color or stencil situation dependant
+    // NOTE: Depth attachment isn't used this way in Vulkan but logically maps to WGPU docs. If
+    // issues, investigate here.
+    if (any((TextureUsage::COLOR_ATTACHMENT | TextureUsage::STENCIL_ATTACHMENT |
+                    TextureUsage::DEPTH_ATTACHMENT) &
+                fUsage)) {
+        retUsage |= wgpu::TextureUsage::RenderAttachment;
+    }
+
+    // This is from Vulkan logic- if there are any issues try disabling this first, allows perf
+    // benefit though
+    const bool useTransientAttachment =
+            // Usage consists of attachment flags only.
+            none(fUsage & ~TextureUsage::ALL_ATTACHMENTS) &&
+            // Usage contains at least one attachment flag.
+            any(fUsage & TextureUsage::ALL_ATTACHMENTS) &&
+            // Depth resolve cannot use transient attachment because it uses a custom shader.
+            // TODO: see VulkanDriver::isDepthStencilResolveSupported() to know when to remove this
+            // restriction.
+            // Note that the custom shader does not resolve stencil. We do need to move to vk 1.2
+            // and above to be able to support stencil resolve (along with depth).
+            !(any(fUsage & TextureUsage::DEPTH_ATTACHMENT) && samples > 1);
+    if (useTransientAttachment) {
+        retUsage |= wgpu::TextureUsage::TransientAttachment;
+    }
+    // NOTE: Unused wgpu flags:
+    //  StorageAttachment
+
+    // NOTE: Unused Filament flags:
+    //  SUBPASS_INPUT VK goes to input attachment which we don't support right now
+    //  PROTECTED
+    return retUsage;
+}
+
 [[nodiscard]] wgpu::TextureAspect fToWGPUTextureViewAspect(TextureUsage const& fUsage,
         TextureFormat const& fFormat) {
 
@@ -208,17 +286,13 @@ WebGPUTexture::WebGPUTexture(const SamplerType samplerType, const uint8_t levels
       mAspect{ fToWGPUTextureViewAspect(usage, format) },
       mWebGPUUsage{ fToWGPUTextureUsage(usage, samples,
               mMipmapGenerationStrategy == MipmapGenerationStrategy::SPD_COMPUTE_PASS,
-              mMipmapGenerationStrategy == MipmapGenerationStrategy::RENDER_PASS,
-              device.HasFeature(wgpu::FeatureName::TransientAttachments)) },
-      mViewUsage{ fToWGPUTextureUsage(usage, samples, false, false,
-              device.HasFeature(wgpu::FeatureName::TransientAttachments)) },
-      mDimension{ toWebGPUTextureViewDimension(samplerType) },
+              mMipmapGenerationStrategy == MipmapGenerationStrategy::RENDER_PASS) },
+      mViewUsage{ fToWGPUTextureUsage(usage, samples, false, false) },
+      mDimension{toWebGPUTextureViewDimension(samplerType)},
       mBlockWidth{ filament::backend::getBlockWidth(format) },
       mBlockHeight{ filament::backend::getBlockHeight(format) },
       mDefaultMipLevel{ 0 },
-      mDefaultBaseArrayLayer{ 0 },
-      // an undefined mSwizzle means that is not a swizzle view
-      mSwizzle{} {
+      mDefaultBaseArrayLayer{ 0 } {
     assert_invariant(
             samples == 1 ||
             samples == 4 &&
@@ -281,63 +355,21 @@ WebGPUTexture::WebGPUTexture(WebGPUTexture const* src, const uint8_t baseLevel,
       mTexture{ src->mTexture },
       mDefaultMipLevel{ baseLevel },
       mDefaultBaseArrayLayer{ src->mArrayLayerCount },
-      mMsaaSidecarTexture{ src->mMsaaSidecarTexture },
-      mSwizzle{ src->mSwizzle } {
-    const bool swizzleUndefined{ (src->mSwizzle.r == wgpu::ComponentSwizzle::Undefined) &&
-                                 (src->mSwizzle.g == wgpu::ComponentSwizzle::Undefined) &&
-                                 (src->mSwizzle.b == wgpu::ComponentSwizzle::Undefined) &&
-                                 (src->mSwizzle.a == wgpu::ComponentSwizzle::Undefined) };
-    if (swizzleUndefined) {
-        mDefaultTextureView = makeTextureView(mDefaultMipLevel, levelCount, 0,
-                mDefaultBaseArrayLayer, src->getViewDimension());
-    } else {
-        wgpu::TextureComponentSwizzleDescriptor swizzleDesc{};
-        swizzleDesc.swizzle = mSwizzle;
-        const wgpu::TextureViewDescriptor viewDesc{
-            .nextInChain = &swizzleDesc,
-            .label = "swizzled_texture_view",
-            .format = mTexture.GetFormat(),
-            .dimension = src->getViewDimension(),
-            .baseMipLevel = mDefaultMipLevel,
-            .mipLevelCount = levelCount,
-            .baseArrayLayer = 0,
-            .arrayLayerCount = mDefaultBaseArrayLayer,
-        };
-        mDefaultTextureView = mTexture.CreateView(&viewDesc);
-        FILAMENT_CHECK_POSTCONDITION(mDefaultTextureView)
-                << "Failed to create swizzled Texture view";
+      mDefaultTextureView{ makeTextureView(mDefaultMipLevel, levelCount, 0, mDefaultBaseArrayLayer,
+              toWebGPUTextureViewDimension(src->target)) },
+      mMsaaSidecarTexture{ src->mMsaaSidecarTexture } {}
+
+wgpu::Texture const& WebGPUTexture::getMsaaSidecarTexture(const uint8_t sampleCount) const {
+    if (mMsaaSidecarTexture == nullptr) {
+        return mMsaaSidecarTexture; // nullptr (no such sidecar)
     }
-}
-
-WebGPUTexture::WebGPUTexture(const WebGPUTexture* src,
-        const wgpu::TextureComponentSwizzle nextSwizzle) noexcept
-    : HwTexture{ src->target, src->levels, src->samples, src->width, src->height, src->depth,
-              src->format, src->usage},
-      mViewFormat{ src->mViewFormat },
-      mMipmapGenerationStrategy{ src->mMipmapGenerationStrategy },
-      mWebGPUFormat{ src->mWebGPUFormat },
-      mAspect{ src->mAspect },
-      mWebGPUUsage{ src->mWebGPUUsage },
-      mViewUsage{ src->mViewUsage },
-      mBlockWidth{ src->mBlockWidth },
-      mBlockHeight{ src->mBlockHeight },
-      mArrayLayerCount{ src->mArrayLayerCount },
-      mTexture{ src->mTexture },
-      mDefaultMipLevel{ src->mDefaultMipLevel },
-      mDefaultBaseArrayLayer{ 0 },
-      mMsaaSidecarTexture{src->mMsaaSidecarTexture},
-      mSwizzle{ composeSwizzle(src->getSwizzle(), nextSwizzle) } {
-    wgpu::TextureComponentSwizzleDescriptor swizzleDesc{};
-    swizzleDesc.swizzle = mSwizzle;
-
-    const wgpu::TextureViewDescriptor viewDesc{
-        .nextInChain = &swizzleDesc,
-        .label = "swizzled_texture_view",
-        .format = mTexture.GetFormat(),
-        .dimension = src->getViewDimension(),
-    };
-    mDefaultTextureView = mTexture.CreateView(&viewDesc);
-    FILAMENT_CHECK_POSTCONDITION(mDefaultTextureView) << "Failed to create swizzled Texture view";
+    FILAMENT_CHECK_PRECONDITION(sampleCount == mMsaaSidecarTexture.GetSampleCount())
+            << "The MSAA sidecar texture has a different sample count ("
+            << mMsaaSidecarTexture.GetSampleCount() << ") than requested (" << +sampleCount
+            << "). Note that this restriction was written when WebGPU only supported msaa "
+               "textures with 4 samples. If that has changed, this implementation should be "
+               "updated (e.g. map of sidecar textures by sampleCount or something).";
+    return mMsaaSidecarTexture;
 }
 
 bool WebGPUTexture::supportsMultipleMipLevelsViaStorageBinding(const wgpu::TextureFormat format) {

filament/backend/src/webgpu/WebGPUTexture.h

@@ -34,25 +34,11 @@ public:
         NONE,
     };
 
-    /**
-     * Creates a Filament texture and a texture view
-     */
     WebGPUTexture(SamplerType, uint8_t levels, TextureFormat, uint8_t samples, uint32_t width,
             uint32_t height, uint32_t depth, TextureUsage, wgpu::Device const&) noexcept;
 
-    /**
-     * Creates a "Filament Texture View", where the underlying texture is the same as the source,
-     * but the view elements differ
-     */
     WebGPUTexture(WebGPUTexture const* src, uint8_t baseLevel, uint8_t levelCount) noexcept;
 
-    /**
-     * @param nextSwizzle the swizzle to apply to the source texture.
-     * Creates a new texture view by composing the nextSwizzle with the source which itself it
-     * could be a swizzled view
-     */
-    WebGPUTexture(WebGPUTexture const* source, wgpu::TextureComponentSwizzle nextSwizzle) noexcept;
-
     [[nodiscard]] wgpu::TextureAspect getAspect() const { return mAspect; }
 
     [[nodiscard]] size_t getBlockWidth() const { return mBlockWidth; }
@@ -61,6 +47,8 @@ public:
 
     [[nodiscard]] wgpu::Texture const& getTexture() const { return mTexture; }
 
+    [[nodiscard]] wgpu::Texture const& getMsaaSidecarTexture(uint8_t sampleCount) const;
+
     [[nodiscard]] wgpu::TextureView const& getDefaultTextureView() const {
         return mDefaultTextureView;
     }
@@ -71,8 +59,6 @@ public:
     [[nodiscard]] wgpu::TextureViewDimension getViewDimension() const { return mDimension; }
     [[nodiscard]] wgpu::TextureFormat getViewFormat() const { return mViewFormat; }
 
-    [[nodiscard]] wgpu::TextureComponentSwizzle getSwizzle() const { return mSwizzle; }
-
     [[nodiscard]] uint32_t getArrayLayerCount() const { return mArrayLayerCount; }
 
     [[nodiscard]] MipmapGenerationStrategy getMipmapGenerationStrategy() const {
@@ -92,15 +78,21 @@ public:
 
     /**
      * @param samples The number of samples the underlying texture supports
-     * @param mipLevel The mip level into the underlying texture for which this view will reference
+     * @param mipLevel The mip level into the underyling texture for which this view will reference
      * (this view will only have one mip level)
-     * @param arrayLayer The layer into the underlying texture for which this view will reference
+     * @param arrayLayer The layer into the underyling texture for which this view will reference
      * (this view will only have one layer)
      * @return A texture view for the MSAA sidecar texture
      */
     wgpu::TextureView makeMsaaSidecarTextureViewIfTextureSidecarExists(uint8_t samples,
             uint8_t mipLevel, uint32_t arrayLayer) const;
 
+    /**
+     * @return nullptr if a MSAA sidecar texture is not appliable, otherwise a view to one
+     */
+    [[nodiscard]] wgpu::TextureView makeMsaaSidecarTextureView(wgpu::Texture const&,
+            uint8_t mipLevel, uint32_t arrayLayer) const;
+
     /**
      * @param format a required texture format (can be a view to a different underlying texture
      * format)
@@ -137,8 +129,6 @@ private:
     // by sampleCount or something like that.
     // For now that complexity and cost is not warranted due to WebGPU's restrictions.
     wgpu::Texture mMsaaSidecarTexture = nullptr;
-    // an undefined mSwizzle means that is not a swizzle view
-    wgpu::TextureComponentSwizzle mSwizzle{};
 
     [[nodiscard]] wgpu::TextureView makeTextureView(const uint8_t& baseLevel,
             const uint8_t& levelCount, const uint32_t& baseArrayLayer,

filament/backend/src/webgpu/WebGPUTextureHelpers.h

@@ -198,21 +198,81 @@ namespace filament::backend {
         case TextureFormat::RGB_BPTC_SIGNED_FLOAT:   return wgpu::TextureFormat::BC6HRGBFloat;
         case TextureFormat::RGBA_BPTC_UNORM:         return wgpu::TextureFormat::BC7RGBAUnorm;
         case TextureFormat::SRGB_ALPHA_BPTC_UNORM:   return wgpu::TextureFormat::BC7RGBAUnormSrgb;
-        case TextureFormat::RGB9_E5:                 return wgpu::TextureFormat::RGB9E5Ufloat;
-        case TextureFormat::RGB8:                    return wgpu::TextureFormat::RGBA8Unorm;
-        case TextureFormat::SRGB8:                   return wgpu::TextureFormat::RGBA8UnormSrgb;
-        case TextureFormat::RGB8_SNORM:              return wgpu::TextureFormat::RGBA8Snorm;
-        case TextureFormat::RGB8UI:                  return wgpu::TextureFormat::RGBA8Uint;
-        case TextureFormat::RGB8I:                   return wgpu::TextureFormat::RGBA8Sint;
+        case TextureFormat::RGB565:
+            // No direct mapping in wgpu. Could potentially map to RGBA8Unorm
+            // and discard the alpha and lower precision.
+            FWGPU_LOGW << "Requested Filament texture format RGB565 but getting "
+                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
+            return wgpu::TextureFormat::Undefined;
+        case TextureFormat::RGB9_E5: return wgpu::TextureFormat::RGB9E5Ufloat;
+        case TextureFormat::RGB5_A1:
+            // No direct mapping in wgpu. Could potentially map to RGBA8Unorm
+            // and handle the packing/unpacking in shaders.
+            FWGPU_LOGW << "Requested Filament texture format RGB5_A1 but getting "
+                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
+            return wgpu::TextureFormat::Undefined;
+        case TextureFormat::RGBA4:
+            // No direct mapping in wgpu. Could potentially map to RGBA8Unorm
+            // and handle the packing/unpacking in shaders.
+            FWGPU_LOGW << "Requested Filament texture format RGBA4 but getting "
+                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
+            return wgpu::TextureFormat::Undefined;
+        case TextureFormat::RGB8:
+            FWGPU_LOGW << "Requested Filament texture format RGB8 but getting "
+                          "wgpu::TextureFormat::RGBA8Unorm (no direct sRGB equivalent in wgpu "
+                          "without alpha)";
+            return wgpu::TextureFormat::RGBA8Unorm;
+        case TextureFormat::SRGB8:
+            FWGPU_LOGW << "Requested Filament texture format SRGB8 but getting "
+                          "wgpu::TextureFormat::RGBA8UnormSrgb (no direct sRGB equivalent in wgpu "
+                          "without alpha)";
+            return wgpu::TextureFormat::RGBA8UnormSrgb;
+        case TextureFormat::RGB8_SNORM:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB8_SNORM but getting "
+                       "wgpu::TextureFormat::RGBA8Snorm (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA8Snorm;
+        case TextureFormat::RGB8UI:
+            FWGPU_LOGW << "Requested Filament texture format RGB8UI but getting "
+                          "wgpu::TextureFormat::RGBA8Uint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA8Uint;
+        case TextureFormat::RGB8I:
+            FWGPU_LOGW << "Requested Filament texture format RGB8I but getting "
+                          "wgpu::TextureFormat::RGBA8Sint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA8Sint;
         case TextureFormat::R11F_G11F_B10F:          return wgpu::TextureFormat::RG11B10Ufloat;
         case TextureFormat::UNUSED:                  return wgpu::TextureFormat::Undefined;
         case TextureFormat::RGB10_A2:                return wgpu::TextureFormat::RGB10A2Unorm;
-        case TextureFormat::RGB16F:                  return wgpu::TextureFormat::RGBA16Float;
-        case TextureFormat::RGB16UI:                 return wgpu::TextureFormat::RGBA16Uint;
-        case TextureFormat::RGB16I:                  return wgpu::TextureFormat::RGBA16Sint;
-        case TextureFormat::RGB32F:                  return wgpu::TextureFormat::RGBA32Float;
-        case TextureFormat::RGB32UI:                 return wgpu::TextureFormat::RGBA32Uint;
-        case TextureFormat::RGB32I:                  return wgpu::TextureFormat::RGBA32Sint;
+        case TextureFormat::RGB16F:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB16F but getting "
+                       "wgpu::TextureFormat::RGBA16Float (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA16Float;
+        case TextureFormat::RGB16UI:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB16UI but getting "
+                       "wgpu::TextureFormat::RGBA16Uint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA16Uint;
+        case TextureFormat::RGB16I:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB16I but getting "
+                       "wgpu::TextureFormat::RGBA16Sint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA16Sint;
+        case TextureFormat::RGB32F:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB32F but getting "
+                       "wgpu::TextureFormat::RGBA32Float (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA32Float;
+        case TextureFormat::RGB32UI:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB32UI but getting "
+                       "wgpu::TextureFormat::RGBA32Uint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA32Uint;
+        case TextureFormat::RGB32I:
+            FWGPU_LOGW
+                    << "Requested Filament texture format RGB32I but getting "
+                       "wgpu::TextureFormat::RGBA32Sint (no direct mapping in wgpu without alpha)";
+            return wgpu::TextureFormat::RGBA32Sint;
         case TextureFormat::DXT1_RGB:                return wgpu::TextureFormat::BC1RGBAUnorm;
         case TextureFormat::DXT1_RGBA:               return wgpu::TextureFormat::BC1RGBAUnorm;
         case TextureFormat::DXT3_RGBA:               return wgpu::TextureFormat::BC2RGBAUnorm;
@@ -221,20 +281,6 @@ namespace filament::backend {
         case TextureFormat::DXT1_SRGBA:              return wgpu::TextureFormat::BC1RGBAUnormSrgb;
         case TextureFormat::DXT3_SRGBA:              return wgpu::TextureFormat::BC2RGBAUnormSrgb;
         case TextureFormat::DXT5_SRGBA:              return wgpu::TextureFormat::BC3RGBAUnormSrgb;
-        // No direct mapping in wgpu. Could potentially map to RGBA8Unorm and handle the
-        // packing/unpacking in shaders.
-        case TextureFormat::RGB565:
-            FWGPU_LOGW << "Requested Filament texture format RGB565 but getting "
-                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
-            return wgpu::TextureFormat::Undefined;
-        case TextureFormat::RGB5_A1:
-            FWGPU_LOGW << "Requested Filament texture format RGB5_A1 but getting "
-                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
-            return wgpu::TextureFormat::Undefined;
-        case TextureFormat::RGBA4:
-            FWGPU_LOGW << "Requested Filament texture format RGBA4 but getting "
-                          "wgpu::TextureFormat::Undefined (no direct mapping in wgpu)";
-            return wgpu::TextureFormat::Undefined;
     }
 }
 
@@ -295,20 +341,6 @@ namespace filament::backend {
     }
 }
 
-[[nodiscard]] constexpr wgpu::ComponentSwizzle toWGPUComponentSwizzle(const backend::TextureSwizzle swizzle) {
-    switch (swizzle) {
-        case backend::TextureSwizzle::CHANNEL_0:       return wgpu::ComponentSwizzle::R;
-        case backend::TextureSwizzle::CHANNEL_1:       return wgpu::ComponentSwizzle::G;
-        case backend::TextureSwizzle::CHANNEL_2:       return wgpu::ComponentSwizzle::B;
-        case backend::TextureSwizzle::CHANNEL_3:       return wgpu::ComponentSwizzle::A;
-        case backend::TextureSwizzle::SUBSTITUTE_ZERO: return wgpu::ComponentSwizzle::Zero;
-        case backend::TextureSwizzle::SUBSTITUTE_ONE:  return wgpu::ComponentSwizzle::One;
-        default:
-             FWGPU_LOGW << "invalid swizzle component returning wgpu::ComponentSwizzle::Undefined";
-            return wgpu::ComponentSwizzle::Undefined;
-    }
-}
-
 /**
  * @return true if https://www.w3.org/TR/webgpu/#copy-compatible which states:
  *  Two GPUTextureFormats format1 and format2 are copy-compatible if:
@@ -502,188 +534,6 @@ namespace filament::backend {
     }
 }
 
-/**
- * Todo: should this take into account sRGB/linear when determining if conversion is necessary?
- *       For instance, if the output format is the same as the input, except the output is sRGB
- *       do we really need to do a conversion? If the answer is no, we should do that check here
- *       and return conversionNecessary false in that case.
- *       However, doing a straight-forward comparison is the safest most conservative thing to do
- *       for functional correctness and NOT doing a conversion in such cases could be considered
- *       an optimization. Thus, consider the optimization when we have better test coverage to
- *       experiment with such a refactor.
- * @return True if theres a format mismatch
- */
-[[nodiscard]] constexpr bool conversionNecessary(const wgpu::TextureFormat source,
-        const wgpu::TextureFormat destination, const PixelDataType pixelDataType) {
-    return source != destination && pixelDataType != PixelDataType::COMPRESSED;
-}
-
-/**
- * @param fUsage Filament's requested texture usage
- * @param samples How many samples to use for MSAA
- * @param needsComputeStorageSupport if we need to use this texture as storage binding in something
- *                                   like a compute shader
- * @param needsRenderAttachmentSupport if we need to use this texture as a render pass attachment
- *                                     in something like a render pass blit (e.g. mipmap generation)
- * @param deviceSupportsTransientAttachments if the device itself supports Render Attachments
- * @return The appropriate texture usage flags for the underlying texture
- */
-[[nodiscard]] constexpr wgpu::TextureUsage fToWGPUTextureUsage(TextureUsage const& fUsage,
-        const uint8_t samples, const bool needsComputeStorageSupport,
-        const bool needsRenderAttachmentSupport, const bool deviceSupportsTransientAttachments) {
-    wgpu::TextureUsage retUsage{ wgpu::TextureUsage::None };
-
-    if (any(TextureUsage::BLIT_SRC & fUsage)) {
-        retUsage |= (wgpu::TextureUsage::CopySrc | wgpu::TextureUsage::TextureBinding);
-    }
-    if (any(TextureUsage::BLIT_DST & fUsage)) {
-        retUsage |= (wgpu::TextureUsage::CopyDst | wgpu::TextureUsage::RenderAttachment);
-    }
-    if (any(TextureUsage::UPLOADABLE & fUsage)) {
-        retUsage |= wgpu::TextureUsage::CopyDst;
-    }
-    if (any(TextureUsage::GEN_MIPMAPPABLE & fUsage)) {
-        retUsage |= (wgpu::TextureUsage::RenderAttachment | wgpu::TextureUsage::TextureBinding);
-    }
-    if (any(TextureUsage::SAMPLEABLE & fUsage)) {
-        retUsage |= wgpu::TextureUsage::TextureBinding;
-    }
-    // if needsComputeStorageSupport we need to read and write to the texture in a shader and thus
-    // require CopySrc, CopyDst, TextureBinding, & StorageBinding
-    if (needsComputeStorageSupport) {
-        retUsage |= (wgpu::TextureUsage::StorageBinding |
-                     wgpu::TextureUsage::CopySrc |
-                     wgpu::TextureUsage::CopyDst |
-                     wgpu::TextureUsage::TextureBinding);
-    }
-
-    wgpu::TextureUsage transientAttachmentNeeded{ wgpu::TextureUsage::None };
-    const bool useTransientAttachment {
-            deviceSupportsTransientAttachments &&
-            // Usage consists of attachment flags only.
-            none(fUsage & ~TextureUsage::ALL_ATTACHMENTS) &&
-            // Usage contains at least one attachment flag.
-            any(fUsage & TextureUsage::ALL_ATTACHMENTS) &&
-            // Depth resolve cannot use transient attachment because it uses a custom shader.
-            // TODO: see VulkanDriver::isDepthStencilResolveSupported() to know when to remove this
-            // restriction.
-            // Note that the custom shader does not resolve stencil. We do need to move to vk 1.2
-            // and above to be able to support stencil resolve (along with depth).
-            !(any(fUsage & TextureUsage::DEPTH_ATTACHMENT) && samples > 1)};
-    if (useTransientAttachment) {
-        transientAttachmentNeeded |= wgpu::TextureUsage::TransientAttachment;
-    }
-
-    // A texture that is a blit destination or render attachment will often need to be
-    // a copy source for subsequent operations (e.g., mipmap generation, readbacks).
-    // However, we dont need to add the CopySrc IF its a transientAttachment
-    if (any((TextureUsage::BLIT_DST | TextureUsage::COLOR_ATTACHMENT |
-                    TextureUsage::DEPTH_ATTACHMENT) &
-                fUsage)) {
-        if (!useTransientAttachment) {
-            retUsage |= wgpu::TextureUsage::CopySrc;
-        }
-    }
-
-    if (needsRenderAttachmentSupport) {
-        retUsage |= wgpu::TextureUsage::RenderAttachment;
-    }
-    // WGPU Render attachment covers either color or stencil situation dependant
-    if (any((TextureUsage::COLOR_ATTACHMENT | TextureUsage::STENCIL_ATTACHMENT |
-                    TextureUsage::DEPTH_ATTACHMENT) &
-                fUsage)) {
-        retUsage |= wgpu::TextureUsage::RenderAttachment;
-        retUsage |= transientAttachmentNeeded;
-    }
-
-    // NOTE: Unused wgpu flags:
-    //  StorageAttachment
-
-    // NOTE: Unused Filament flags:
-    //  SUBPASS_INPUT: VK goes to input attachment which we don't support right now
-    //  PROTECTED
-    return retUsage;
-}
-
-/**
- * "Updates" the previous swizzle parameters with the next/new parameters, essentially stacking
- * swizzle parameters. Essentially this combines the parameters such that the result is
- * the same mapping of channels you would get if you first applied the "previous" mapping
- * followed by the "next" mapping.
- */
-[[nodiscard]] constexpr wgpu::TextureComponentSwizzle composeSwizzle(
-        wgpu::TextureComponentSwizzle const& previous, wgpu::TextureComponentSwizzle const& next) {
-    // start with the "next" values...
-    wgpu::ComponentSwizzle result[] { next.r, next.g, next.b, next.a };
-    // track the current channel as we go through values to know what to assign in the undefined
-    // case...
-    wgpu::ComponentSwizzle channel{wgpu::ComponentSwizzle::R};
-    for (auto& out: result) {
-        switch (out) {
-            case wgpu::ComponentSwizzle::R:
-                if (previous.r != wgpu::ComponentSwizzle::Undefined) {
-                    out = previous.r;
-                }
-                break;
-            case wgpu::ComponentSwizzle::G:
-                if (previous.g != wgpu::ComponentSwizzle::Undefined) {
-                    out = previous.g;
-                }
-                break;
-            case wgpu::ComponentSwizzle::B:
-                if (previous.b != wgpu::ComponentSwizzle::Undefined) {
-                    out = previous.b;
-                }
-                break;
-            case wgpu::ComponentSwizzle::A:
-                if (previous.a != wgpu::ComponentSwizzle::Undefined) {
-                    out = previous.a;
-                }
-                break;
-            // Do not modify the (next) value for explicit One or Zero...
-            case wgpu::ComponentSwizzle::Zero:
-            case wgpu::ComponentSwizzle::One:
-                break;
-            case wgpu::ComponentSwizzle::Undefined:
-                // fallback to the previous value if the next value is undefined...
-                switch (channel) {
-                    case wgpu::ComponentSwizzle::R:
-                        out = previous.r;
-                        break;
-                    case wgpu::ComponentSwizzle::G:
-                        out = previous.g;
-                        break;
-                    case wgpu::ComponentSwizzle::B:
-                        out = previous.b;
-                        break;
-                    case wgpu::ComponentSwizzle::A:
-                        out = previous.a;
-                        break;
-                    default:
-                        // should not be possible...
-                        out = wgpu::ComponentSwizzle::Undefined;
-                        break;
-                }
-                break;
-        }
-        switch (channel) {
-            case wgpu::ComponentSwizzle::R:
-                channel = wgpu::ComponentSwizzle::G;
-                break;
-            case wgpu::ComponentSwizzle::G:
-                channel = wgpu::ComponentSwizzle::B;
-                break;
-            case wgpu::ComponentSwizzle::B:
-                channel = wgpu::ComponentSwizzle::A;
-                break;
-            default:
-                channel = wgpu::ComponentSwizzle::Undefined;
-                break;
-        }
-    }
-    return { .r = result[0], .g = result[1], .b = result[2], .a = result[3] };
-}
-
 } // namespace filament::backend
 
 #endif // TNT_FILAMENT_BACKEND_WEBGPUTEXTUREHELPERS_H

filament/backend/src/webgpu/platform/WebGPUPlatform.cpp

@@ -67,11 +67,9 @@ constexpr std::array REQUIRED_FEATURES = {
 };
 
 constexpr std::array OPTIONAL_FEATURES = {
-    wgpu::FeatureName::CoreFeaturesAndLimits,
     wgpu::FeatureName::DepthClipControl,
     wgpu::FeatureName::Depth32FloatStencil8,
-    wgpu::FeatureName::TextureComponentSwizzle,
-};
+    wgpu::FeatureName::CoreFeaturesAndLimits };
 
 enum class LimitToValidate : uint8_t {
     begin = 0,// needs to be first for iterating through all possible values in the enum
@@ -507,7 +505,6 @@ struct AdapterDetailsHash final {
 
 // selects one preferred adapter or panics if none can be found
 wgpu::Adapter selectPreferredAdapter(
-        WebGPUPlatform::Configuration const& configuration,
         std::unordered_set<AdapterDetails, AdapterDetailsHash> const& compatibleAdapters) {
     // for each unique adapter...
     AdapterDetails const* selectedAdapter = nullptr;
@@ -515,10 +512,6 @@ wgpu::Adapter selectPreferredAdapter(
 
     // choose the most desirable adapter that meets the minimum requirements...
     for (AdapterDetails const& details: compatibleAdapters) {
-        if (configuration.forceBackendType != wgpu::BackendType::Undefined &&
-                configuration.forceBackendType != details.info.backendType) {
-            continue;
-        }
         if (!adapterMeetsMinimumRequirements(details)) {
             continue;
         }
@@ -594,7 +587,7 @@ wgpu::Adapter WebGPUPlatform::requestAdapter(wgpu::Surface const& surface) {
     }
     const std::unordered_set<AdapterDetails, AdapterDetailsHash> compatibleAdapters =
             requestCompatibleAdapters(mInstance, requests);
-    return selectPreferredAdapter(getConfiguration(), compatibleAdapters);
+    return selectPreferredAdapter(compatibleAdapters);
 }
 
 wgpu::Device WebGPUPlatform::requestDevice(wgpu::Adapter const& adapter) {

filament/backend/src/webgpu/platform/WebGPUPlatformApple.mm

@@ -38,12 +38,7 @@ std::vector<wgpu::RequestAdapterOptions> WebGPUPlatform::getAdapterOptions() {
     constexpr std::array powerPreferences = {
         wgpu::PowerPreference::HighPerformance,
         wgpu::PowerPreference::LowPower };
-    constexpr std::array backendTypes = {
-        wgpu::BackendType::Metal,
-        // To enable software rasterization on MacOS, we need to ensure Vulkan adapters are
-        // available.
-        wgpu::BackendType::Vulkan,
-    };
+    constexpr std::array backendTypes = { wgpu::BackendType::Metal };
     constexpr std::array forceFallbackAdapters = { false, true };
     constexpr size_t totalCombinations =
             powerPreferences.size() * backendTypes.size() * forceFallbackAdapters.size();

filament/backend/src/webgpu/platform/WebGPUPlatformLinux.cpp

@@ -41,21 +41,36 @@
         uint32_t height;
     } wl;
     }// namespace
-#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11) && defined(FILAMENT_SUPPORTS_XLIB)
-    // TODO: we should allow for headless on Linux explicitly. Right now headless path doesn't work
+#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11)
+    // TODO: we should allow for headless on Linux explicitly. Right now this is the headless path
+    // (with no FILAMENT_SUPPORTS_XCB or FILAMENT_SUPPORTS_XLIB).
     #include <dlfcn.h>
-    #include <X11/Xlib.h>
-
-    static constexpr const char* LIBRARY_X11 { "libX11.so.6" };
-
-    namespace {
+    #if defined(FILAMENT_SUPPORTS_XCB)
+        #include <xcb/xcb.h>
+        namespace {
+        typedef xcb_connection_t* (*XCB_CONNECT)(const char* displayname, int* screenp);
+        }// namespace
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XLIB)
+        #include <X11/Xlib.h>
+        namespace {
         typedef Display* (*X11_OPEN_DISPLAY)(const char*);
-        struct XEnv final {
-            X11_OPEN_DISPLAY openDisplay;
-            Display* display { nullptr };
-            void* library { nullptr };
-        } g_x11;
-    } // namespace
+        }// namespace
+    #endif
+    static constexpr const char* LIBRARY_X11 = "libX11.so.6";
+    namespace {
+    struct XEnv {
+    #if defined(FILAMENT_SUPPORTS_XCB)
+        XCB_CONNECT xcbConnect;
+        xcb_connection_t* connection;
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XLIB)
+        X11_OPEN_DISPLAY openDisplay;
+        Display* display;
+    #endif
+        void* library = nullptr;
+    } g_x11;
+    }// namespace
 #else
     #error Not a supported Linux or FeeBSD + WebGPU platform
 #endif
@@ -101,26 +116,59 @@ wgpu::Extent2D WebGPUPlatform::getSurfaceExtent(void* nativeWindow) const {
     surfaceExtent.height = ptrval->height;
     FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
             << "Unable to get window size for Linux Wayland-backed surface.";
-#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11) && defined (FILAMENT_SUPPORTS_XLIB)
+#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11)
     if (g_x11.library == nullptr) {
         g_x11.library = dlopen(LIBRARY_X11, RTLD_LOCAL | RTLD_NOW);
         FILAMENT_CHECK_PRECONDITION(g_x11.library) << "Unable to open X11 library.";
+        #if defined(FILAMENT_SUPPORTS_XCB)
+            g_x11.xcbConnect = (XCB_CONNECT) dlsym(g_x11.library, "xcb_connect");
+            int screen = 0;
+            g_x11.connection = g_x11.xcbConnect(nullptr, &screen);
+        #endif
+        #if defined(FILAMENT_SUPPORTS_XLIB)
+            g_x11.openDisplay = (X11_OPEN_DISPLAY) dlsym(g_x11.library, "XOpenDisplay");
+            g_x11.display = g_x11.openDisplay(NULL);
+            FILAMENT_CHECK_PRECONDITION(g_x11.display) << "Unable to open X11 display.";
+        #endif
     }
-    if (g_x11.openDisplay == nullptr) {
-        g_x11.openDisplay = reinterpret_cast<X11_OPEN_DISPLAY>(dlsym(g_x11.library, "XOpenDisplay"));
-        g_x11.display = g_x11.openDisplay(NULL);
-        FILAMENT_CHECK_PRECONDITION(g_x11.display) << "Unable to open X11 display.";
-    }
-    Window window { reinterpret_cast<Window const>(nativeWindow) };
-    XWindowAttributes windowAttributes;
-    Status ok { XGetWindowAttributes(g_x11.display, window, &windowAttributes) };
-    FILAMENT_CHECK_PRECONDITION(ok != 0) << " XGetWindowAttributes failed";
-    surfaceExtent.width = static_cast<uint32_t>(windowAttributes.width);
-    surfaceExtent.height = static_cast<uint32_t>(windowAttributes.height);
-
+    #if defined(FILAMENT_SUPPORTS_XCB) || defined(FILAMENT_SUPPORTS_XLIB)
+        bool useXcb = false;
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XCB)
+        #if defined(FILAMENT_SUPPORTS_XLIB)
+            useXcb = (SWAP_CHAIN_CONFIG_ENABLE_XCB) != 0;
+        #else
+            useXcb = true;
+        #endif
+        if (useXcb) {
+            const xcb_setup_t* setup = xcb_get_setup(g_x11.connection);
+            xcb_screen_iterator_t screen_iter = xcb_setup_roots_iterator(setup);
+            xcb_screen_t* screen = screen_iter.data;
+            surfaceExtent.width = static_cast<uint32_t>(screen->width_in_pixels);
+            surfaceExtent.height = static_cast<uint32_t>(screen->height_in_pixels);
+            FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
+                    << "Unable to get window surface size for Linux (or FreeBSD) "
+                       "XCB-backed surface.";
+        }
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XLIB)
+        if (!useXcb) {
+            int screenNumber = DefaultScreen(g_x11.display);
+            Screen* screen = ScreenOfDisplay(g_x11.display, screenNumber);
+            surfaceExtent.width = static_cast<uint32_t>(WidthOfScreen(screen));
+            surfaceExtent.height = static_cast<uint32_t>(HeightOfScreen(screen));
+            FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
+                    << "Unable to get window surface size for Linux (or FreeBSD) "
+                       "XLib-backed surface.";
+        }
+    #endif
     FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
             << "Cannot get window surface size for X11 surface for Linux (or FreeBSD) OS "
-               "(not built with support for XLIB?)";
+               "(not built with support for XCB or XLIB?)";
+#elif defined(__linux__)
+    FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
+            << "Cannot get window surface size for Linux (or FreeBSD) OS "
+               "(not built with support for Wayland or X11?)";
 #else
     FILAMENT_CHECK_POSTCONDITION(surfaceExtent.width != 0 && surfaceExtent.height != 0)
             << "Not a supported (Linux) OS + WebGPU platform";
@@ -128,40 +176,80 @@ wgpu::Extent2D WebGPUPlatform::getSurfaceExtent(void* nativeWindow) const {
     return surfaceExtent;
 }
 
-wgpu::Surface WebGPUPlatform::createSurface(void* nativeWindow, uint64_t /*flags*/) {
+wgpu::Surface WebGPUPlatform::createSurface(void* nativeWindow, uint64_t flags) {
     wgpu::Surface surface = nullptr;
 #if defined(__linux__) && defined(FILAMENT_SUPPORTS_WAYLAND)
     wl* ptrval = reinterpret_cast<wl*>(nativeWindow);
     wgpu::SurfaceSourceWaylandSurface surfaceSourceWayland{};
     surfaceSourceWayland.display = ptrval->display;
     surfaceSourceWayland.surface = ptrval->surface;
-    wgpu::SurfaceDescriptor const surfaceDescriptor {
+    wgpu::SurfaceDescriptor surfaceDescriptor{
         .nextInChain = &surfaceSourceWayland,
         .label = "linux_wayland_surface"
     };
     surface = mInstance.CreateSurface(&surfaceDescriptor);
     FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
             << "Unable to create Linux Wayland-backed surface.";
-#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11) && defined(FILAMENT_SUPPORTS_XLIB)
+#elif defined(LINUX_OR_FREEBSD) && defined(FILAMENT_SUPPORTS_X11)
     if (g_x11.library == nullptr) {
         g_x11.library = dlopen(LIBRARY_X11, RTLD_LOCAL | RTLD_NOW);
         FILAMENT_CHECK_PRECONDITION(g_x11.library) << "Unable to open X11 library.";
+        #if defined(FILAMENT_SUPPORTS_XCB)
+            g_x11.xcbConnect = (XCB_CONNECT) dlsym(g_x11.library, "xcb_connect");
+            int screen = 0;
+            g_x11.connection = g_x11.xcbConnect(nullptr, &screen);
+        #endif
+        #if defined(FILAMENT_SUPPORTS_XLIB)
+            g_x11.openDisplay = (X11_OPEN_DISPLAY) dlsym(g_x11.library, "XOpenDisplay");
+            g_x11.display = g_x11.openDisplay(NULL);
+            FILAMENT_CHECK_PRECONDITION(g_x11.display) << "Unable to open X11 display.";
+        #endif
     }
-    if (g_x11.openDisplay == nullptr) {
-        g_x11.openDisplay = (X11_OPEN_DISPLAY) dlsym(g_x11.library, "XOpenDisplay");
-        g_x11.display = g_x11.openDisplay(NULL);
-        FILAMENT_CHECK_PRECONDITION(g_x11.display) << "Unable to open X11 display.";
-    }
-    wgpu::SurfaceSourceXlibWindow surfaceSourceXlib{};
-    surfaceSourceXlib.display = g_x11.display;
-    surfaceSourceXlib.window = reinterpret_cast<Window const>(nativeWindow);
-    wgpu::SurfaceDescriptor const surfaceDescriptor {
-        .nextInChain = &surfaceSourceXlib,
-        .label = "linux_xlib_surface"
-    };
-    surface = mInstance.CreateSurface(&surfaceDescriptor);
+    #if defined(FILAMENT_SUPPORTS_XCB) || defined(FILAMENT_SUPPORTS_XLIB)
+        bool useXcb = false;
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XCB)
+        #if defined(FILAMENT_SUPPORTS_XLIB)
+            useXcb = (flags & SWAP_CHAIN_CONFIG_ENABLE_XCB) != 0;
+        #else
+            useXcb = true;
+        #endif
+        if (useXcb) {
+            wgpu::SurfaceSourceXCBWindow surfaceSourceXcb{};
+            surfaceSourceXcb.connection = g_x11.connection;
+
+            // TODO: this looks really wrong, please fix!!
+            surfaceSourceXcb.window = *((uint32_t*) nativeWindow);
+            wgpu::SurfaceDescriptor surfaceDescriptor{
+                .nextInChain = &surfaceSourceXcb,
+                .label = "linux_xcb_surface"
+            };
+            surface = mInstance.CreateSurface(&surfaceDescriptor);
+            FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
+                    << "Unable to create Linux (or FreeBSD) XCB-backed surface.";
+        }
+    #endif
+    #if defined(FILAMENT_SUPPORTS_XLIB)
+        if (!useXcb) {
+            wgpu::SurfaceSourceXlibWindow surfaceSourceXlib{};
+            surfaceSourceXlib.display = g_x11.display;
+            surfaceSourceXlib.window = reinterpret_cast<uint64_t>(nativeWindow);
+            wgpu::SurfaceDescriptor surfaceDescriptor{
+                .nextInChain = &surfaceSourceXlib,
+                .label = "linux_xlib_surface"
+            };
+            surface = mInstance.CreateSurface(&surfaceDescriptor);
+            FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
+                    << "Unable to create Linux (or FreeBSD) XLib-backed surface.";
+        }
+    #endif
     FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
-            << "Unable to create Linux (or FreeBSD) XLib-backed surface.";
+            << "Cannot create WebGPU X11 surface for Linux (or FreeBSD) OS "
+               "(not built with support for XCB or XLIB?)";
+#elif defined(__linux__)
+    FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
+            << "Cannot create WebGPU surface for Linux (or FreeBSD) OS "
+               "(not built with support for Wayland or X11?)";
 #else
     FILAMENT_CHECK_POSTCONDITION(surface != nullptr)
             << "Not a supported (Linux) OS + WebGPU platform";

filament/backend/src/webgpu/utils/AsyncTaskCounter.cpp

@@ -1,45 +0,0 @@
-/*
- * Copyright (C) 2025 The Android Open Source Project
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#include "AsyncTaskCounter.h"
-
-#include "utils/debug.h"
-
-#include <mutex>
-
-namespace filament::backend::webgpuutils {
-
-void AsyncTaskCounter::startTask() {
-    std::lock_guard<std::mutex> lock{ mMutex };
-    ++mTasksInProgress;
-}
-
-void AsyncTaskCounter::finishTask() {
-    std::lock_guard<std::mutex> lock{ mMutex };
-    --mTasksInProgress;
-    assert_invariant(mTasksInProgress >= 0);
-    if (mTasksInProgress == 0) {
-        mFinishedCondition.notify_all();
-    }
-}
-
-void AsyncTaskCounter::waitForAllToFinish() {
-    std::unique_lock<std::mutex> lock{ mMutex };
-    mFinishedCondition.wait(lock, [this] { return mTasksInProgress == 0; });
-}
-
-} // namespace filament::backend::webgpuutils
-

filament/backend/src/webgpu/utils/AsyncTaskCounter.h

@@ -1,62 +0,0 @@
-/*
- * Copyright (C) 2025 The Android Open Source Project
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#ifndef TNT_FILAMENT_BACKEND_UTIL_ASYNCWORKCOUNTER_H
-#define TNT_FILAMENT_BACKEND_UTIL_ASYNCWORKCOUNTER_H
-
-#include <condition_variable>
-#include <mutex>
-
-namespace filament::backend::webgpuutils {
-
-/**
- * Counts the number of asynchronous tasks of a given type.
- * This can be used to wait for such a count to decrement to 0 (wait for work to complete) in a
- * thead safe manner.
- */
-class AsyncTaskCounter final {
-public:
-    AsyncTaskCounter() = default;
-    AsyncTaskCounter(AsyncTaskCounter const&) = delete;
-    AsyncTaskCounter(AsyncTaskCounter const&&) = delete;
-    AsyncTaskCounter& operator=(AsyncTaskCounter const&) = delete;
-    AsyncTaskCounter& operator=(AsyncTaskCounter const&&) = delete;
-    ~AsyncTaskCounter() = default;
-
-    /**
-     * Asynchronously (thread-safe) increase the work counter by one
-     */
-    void startTask();
-
-    /**
-     * Asynchronously (thread-safe) decrease the work counter by one
-     */
-    void finishTask();
-
-    /**
-     * Wait for all tasks to finish (for the work counter to reach 0)  (thread-safe)
-     */
-    void waitForAllToFinish();
-
-private:
-    std::mutex mMutex;
-    std::condition_variable mFinishedCondition;
-    int mTasksInProgress{ 0 };
-};
-
-} // namespace filament::backend::webgpuutils
-
-#endif // TNT_FILAMENT_BACKEND_UTIL_ASYNCWORKCOUNTER_H

filament/backend/test/Arguments.cpp

@@ -49,7 +49,7 @@ Backend parseArgumentsForBackend(int argc, char* argv[]) {
                 } else if (arg == "webgpu") {
                     backend = Backend::WEBGPU;
                 } else {
-                    std::cerr << "Unrecognized target API. Must be 'opengl'|'vulkan'|'metal'|'webgpu'."
+                    std::cerr << "Unrecognized target API. Must be 'opengl'|'vulkan'|'metal'."
                               << std::endl
                               << "Defaulting to OpenGL."
                               << std::endl;

filament/backend/test/ImageExpectations.cpp

@@ -85,18 +85,6 @@ std::filesystem::path ScreenshotParams::expectedFilePath() const {
     return expectedDirectoryPath().append(expectedFileName());
 }
 
-std::filesystem::path ScreenshotParams::diffFilePath() const {
-    return diffDirectoryPath().append(diffFileName());
-}
-
-std::filesystem::path ScreenshotParams::diffDirectoryPath() {
-    return BackendTest::binaryDirectory().append("images/diff_images");
-}
-
-std::string ScreenshotParams::diffFileName() const {
-    return absl::StrFormat("%s_diff.png", mFileName);
-}
-
 const std::string ScreenshotParams::filePrefix() const {
     // TODO(b/422804941): If there are platform specific goldens, when on those platforms append a
     //  unique platform identifying string to this.
@@ -140,6 +128,8 @@ void ImageExpectation::compareImage() const {
     if (bytesFilled) {
 #ifndef FILAMENT_IOS
         LoadedPng loadedImage(mParams.expectedFilePath());
+        // Bytewise compare.
+        EXPECT_EQ(loadedImage.bytes().size(), mResult.bytes().size());
         if (loadedImage.bytes().size() != mResult.bytes().size()) {
             // Something is wrong with the size of the expected result, which usually means the file
             // is missing. Fail the test and early return so later steps can assume the image is
@@ -147,41 +137,18 @@ void ImageExpectation::compareImage() const {
             BackendTest::markImageAsFailure(mParams.filePrefix());
             return;
         }
-
-        // Initialize the vector to full saturation on all channels, bad pixels will have
-        // color channels but not alpha set to 0 to produce highly contrasting black pixels.
-        std::vector<unsigned char> imageDiff( loadedImage.bytes().size(), 255 );
         int pixelDeviations = 0;
-        for (int i = 0; i < mResult.bytes().size(); i += 4) {
-            // In order to handle color channels propoerly for the output, stride 4 bytes at a time.
-            // A failure of any byte in a pixel counts as a failure of the whole pixel.
-            for (int j = 0; j < 4; ++j) {
-                if( std::abs( mResult.bytes()[i+j] - loadedImage.bytes()[i+j] ) >
-                    mParams.pixelMatchThreshold() ) {
-                    pixelDeviations++;
-                    imageDiff[i] = 0;
-                    imageDiff[i+1] = 0;
-                    imageDiff[i+2] = 0;
-                    break;
-                }
+        for (int i = 0; i < mResult.bytes().size(); ++i) {
+            if (std::abs( mResult.bytes()[i] - loadedImage.bytes()[i] ) >
+                mParams.pixelMatchThreshold()) {
+                pixelDeviations++;
             }
         }
         EXPECT_LE(pixelDeviations, mParams.allowedPixelDeviations());
         if (pixelDeviations > mParams.allowedPixelDeviations()) {
             BackendTest::markImageAsFailure(mParams.filePrefix());
-            image::LinearImage image;
-            image = image::toLinearWithAlpha<uint8_t>(
-                    mParams.width(), mParams.height(),
-                    mParams.width() * 4, (uint8_t *)imageDiff.data(),
-                    [](uint8_t value) -> float { return value; },
-                    [](filament::math::float4 rgba) -> filament::math::float4 { return rgba; });
-            std::string filePath = mParams.diffFilePath();
-            std::ofstream pngStream(filePath, std::ios::binary | std::ios::trunc);
-            // To avoid going from linear -> sRGB -> linear save the PNG as linear.
-            image::ImageEncoder::encode(pngStream, image::ImageEncoder::Format::PNG_LINEAR, image, "",
-                    filePath);
         }
-        EXPECT_LE(pixelDeviations, mParams.allowedPixelDeviations());
+        // TODO: Add better debug output, such as generating a diff image.
 #else
         // For builds that can't load PNGs (currently iOS only) use the expected hash.
         uint32_t actualHash = mResult.hash();

filament/backend/test/ImageExpectations.h

@@ -58,9 +58,6 @@ public:
     static std::filesystem::path expectedDirectoryPath();
     std::string expectedFileName() const;
     std::filesystem::path expectedFilePath() const;
-    static std::filesystem::path diffDirectoryPath();
-    std::string diffFileName() const;
-    std::filesystem::path diffFilePath() const;
     const std::string filePrefix() const;
     int allowedPixelDeviations() const;
     int pixelMatchThreshold() const;

filament/backend/test/README.md

@@ -19,8 +19,6 @@ code. When cmake is run it will copy this directory to the build output creating
 The unit tests will then write their resulting images into `images/actual_images` in order to make
 the tests easier to debug.
 
-If an image comparison test fails, it writes a diff image to `images/diff_images` where each pixel of the diff is white (255,255,255) if the comparison for the corresponding pixel succeeds and black (0,0,0) of the comparison for the corresponding pixel fails.
-
 ### Python utility for updating golden images and comparing results
 
 Inside the unit test source code directory there is a python script called

filament/backend/test/Workarounds.h

@@ -1,44 +0,0 @@
-/*
- * Copyright (C) 2025 The Android Open Source Project
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#ifndef TNT_BACKEND_TEST_WORKAROUNDS_H
-#define TNT_BACKEND_TEST_WORKAROUNDS_H
-
-#if defined(FILAMENT_SUPPORTS_WEBGPU)
-// consider, if needed, adding (as Vulkan also has transient attachments): || defined(FILAMENT_SUPPORTS_VULKAN)
-// TODO: support TextureUsage::READ_PIXELS (or something like that)
-//        and search-and-replace TEXTURE_USAGE_READ_PIXELS with
-//       "| TextureUsage::READ_PIXELS" when it is available.
-//       Until then, this serves as a workaround for WebGPU testing, such that
-//       it can easily be swapped out for the "right" thing when that is available.
-//       The problem this is trying to address is that the WebGPU backend will
-//       assign a texture with wgpu::TextureUsage::TransientAttachment if the Filament
-//       usage flags don't indicate the pixels are needed CPU-side. If transient attachment is
-//       assigned to the texture, the texture will not be available later for readPixels, which
-//       is often needed for the unit tests. If we suggest that
-//       the texture needs to support being a blit source, then TransientAttachment will
-//       not get assigned. However, this is not quite right in the test, because a test
-//       may not really want to do a blit with the texture, just read the pixels from it later for
-//       image comparison (renderdiff).
-//       Ideally, the test could express this with a TextureUsage::READ_PIXELS flag, but that
-//       does not yet exist. So, a hack would be to explicitly set the BLIT_SRC flag instead in
-//       its place.
-#define TEXTURE_USAGE_READ_PIXELS | TextureUsage::BLIT_SRC
-#else
-#define TEXTURE_USAGE_READ_PIXELS
-#endif
-
-#endif // TNT_BACKEND_TEST_WORKAROUNDS_H

filament/backend/test/test_Blit.cpp

@@ -144,7 +144,6 @@ static void createFaces(DriverApi& dapi, Handle<HwTexture> texture, int baseWidt
 }
 
 TEST_F(BlitTest, ColorMagnify) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     auto& api = getDriverApi();
     mCleanup.addPostCall([&]() { executeCommands(); });
 
@@ -161,18 +160,17 @@ TEST_F(BlitTest, ColorMagnify) {
     api.makeCurrent(swapChain, swapChain);
 
     // Create a source texture.
-    Handle<HwTexture> const srcTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT |
-                    TextureUsage::BLIT_SRC));
+    Handle<HwTexture> const srcTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT));
     const bool flipY = sBackend == Backend::OPENGL;
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 0, float3(0.5, 0, 0), flipY);
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 1, float3(0, 0, 0.5), flipY);
 
     // Create a destination texture.
-    Handle<HwTexture> const dstTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_DST));
+    Handle<HwTexture> const dstTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT));
 
     // Create a RenderTarget for each texture's miplevel.
     Handle<HwRenderTarget> srcRenderTargets[kNumLevels];
@@ -208,7 +206,6 @@ TEST_F(BlitTest, ColorMagnify) {
 }
 
 TEST_F(BlitTest, ColorMinify) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     auto& api = getDriverApi();
     mCleanup.addPostCall([&]() { executeCommands(); });
 
@@ -225,18 +222,17 @@ TEST_F(BlitTest, ColorMinify) {
     api.makeCurrent(swapChain, swapChain);
 
     // Create a source texture.
-    Handle<HwTexture> const srcTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT |
-                    TextureUsage::BLIT_SRC));
+    Handle<HwTexture> const srcTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT));
     const bool flipY = sBackend == Backend::OPENGL;
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 0, float3(0.5, 0, 0), flipY);
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 1, float3(0, 0, 0.5), flipY);
 
     // Create a destination texture.
-    Handle<HwTexture> const dstTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_DST));
+    Handle<HwTexture> const dstTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT));
 
     // Create a RenderTarget for each texture's miplevel.
     Handle<HwRenderTarget> srcRenderTargets[kNumLevels];
@@ -264,7 +260,6 @@ TEST_F(BlitTest, ColorMinify) {
 }
 
 TEST_F(BlitTest, ColorResolve) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     NONFATAL_FAIL_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Nothing is drawn, see b/417229577");
     auto& api = getDriverApi();
@@ -288,12 +283,13 @@ TEST_F(BlitTest, ColorResolve) {
     // Create 4-sample texture.
     Handle<HwTexture> const srcColorTexture = mCleanup.add(api.createTexture(
             SamplerType::SAMPLER_2D, 1, kColorTexFormat, kSampleCount, kSrcTexWidth, kSrcTexHeight,
-            1, TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_SRC | TextureUsage::UPLOADABLE));
+            1,
+            TextureUsage::COLOR_ATTACHMENT));
 
     // Create 1-sample texture.
     Handle<HwTexture> const dstColorTexture = mCleanup.add(api.createTexture(
             SamplerType::SAMPLER_2D, 1, kColorTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_DST));
+            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT));
 
     // Create a 4-sample render target with the 4-sample texture.
     Handle<HwRenderTarget> const srcRenderTarget = mCleanup.add(api.createRenderTarget(
@@ -368,18 +364,18 @@ TEST_F(BlitTest, Blit2DTextureArray) {
     api.makeCurrent(swapChain, swapChain);
 
     // Create a source texture.
-    Handle<HwTexture> srcTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D_ARRAY,
-            kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, kSrcTexDepth,
-            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT |
-                    TextureUsage::BLIT_SRC));
+    Handle<HwTexture> srcTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D_ARRAY, kNumLevels, kSrcTexFormat, 1, kSrcTexWidth,
+            kSrcTexHeight, kSrcTexDepth,
+            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT));
     const bool flipY = sBackend == Backend::OPENGL;
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 0, kSrcTexLayer, float3(0.5, 0, 0),
             flipY);
 
     // Create a destination texture.
-    Handle<HwTexture> dstTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_DST));
+    Handle<HwTexture> dstTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT));
 
     // Create two RenderTargets.
     const int level = 0;
@@ -433,18 +429,17 @@ TEST_F(BlitTest, BlitRegion) {
     api.makeCurrent(swapChain, swapChain);
 
     // Create a source texture.
-    Handle<HwTexture> srcTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT |
-                    TextureUsage::BLIT_SRC));
+    Handle<HwTexture> srcTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kSrcTexFormat, 1, kSrcTexWidth, kSrcTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT));
     const bool flipY = sBackend == Backend::OPENGL;
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 0, float3(0.5, 0, 0), flipY);
     createBitmap(api, srcTexture, kSrcTexWidth, kSrcTexHeight, 1, float3(0, 0, 0.5), flipY);
 
     // Create a destination texture.
-    Handle<HwTexture> dstTexture = mCleanup.add(api.createTexture(SamplerType::SAMPLER_2D,
-            kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
-            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT | TextureUsage::BLIT_DST));
+    Handle<HwTexture> dstTexture = mCleanup.add(api.createTexture(
+            SamplerType::SAMPLER_2D, kNumLevels, kDstTexFormat, 1, kDstTexWidth, kDstTexHeight, 1,
+            TextureUsage::SAMPLEABLE | TextureUsage::COLOR_ATTACHMENT));
 
     // Blit one-quarter of src level 1 to dst level 0.
     Viewport srcRect = {
@@ -492,7 +487,6 @@ TEST_F(BlitTest, BlitRegion) {
 
 TEST_F(BlitTest, BlitRegionToSwapChain) {
     FAIL_IF(Backend::VULKAN, "Crashes due to not finding color attachment, see b/417481493");
-    SKIP_IF(Backend::WEBGPU, "WebGPU Crashes due to not finding color attachment");
     auto& api = getDriverApi();
     mCleanup.addPostCall([&]() { executeCommands(); });
 

filament/backend/test/test_BufferUpdates.cpp

@@ -22,7 +22,6 @@
 #include "SharedShaders.h"
 #include "Skip.h"
 #include "TrianglePrimitive.h"
-#include "Workarounds.h"
 
 namespace test {
 
@@ -156,6 +155,8 @@ TEST_F(BufferUpdatesTest, VertexBufferUpdate) {
 // This test renders two triangles in two separate draw calls. Between the draw calls, a uniform
 // buffer object is partially updated.
 TEST_F(BufferUpdatesTest, BufferObjectUpdateWithOffset) {
+    NONFATAL_FAIL_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
+            "All values including alpha are written as 0, see b/417254943");
 
     auto& api = getDriverApi();
     Cleanup cleanup(api);
@@ -178,9 +179,9 @@ TEST_F(BufferUpdatesTest, BufferObjectUpdateWithOffset) {
     shader.bindUniform<SimpleMaterialParams>(api, ubuffer, kBindingConfig);
 
     // Create a render target.
-    auto colorTexture = cleanup.add(
-            api.createTexture(SamplerType::SAMPLER_2D, 1, TextureFormat::RGBA8, 1, screenWidth(),
-                    screenHeight(), 1, TextureUsage::COLOR_ATTACHMENT TEXTURE_USAGE_READ_PIXELS));
+    auto colorTexture =
+            cleanup.add(api.createTexture(SamplerType::SAMPLER_2D, 1, TextureFormat::RGBA8, 1,
+                    screenWidth(), screenHeight(), 1, TextureUsage::COLOR_ATTACHMENT));
     auto renderTarget = cleanup.add(api.createRenderTarget(TargetBufferFlags::COLOR0, screenWidth(),
             screenHeight(), 1, 0, { { colorTexture } }, {}, {}));
 

filament/backend/test/test_FeedbackLoops.cpp

@@ -104,6 +104,7 @@ TEST_F(BackendTest, FeedbackLoops) {
     SKIP_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::OPENGL),
             "OpenGL image is upside down due to readPixels failing for texture with uploaded image "
             "data");
+    FAIL_IF(Backend::WEBGPU, "BUG");
     auto& api = getDriverApi();
     Cleanup cleanup(api);
 
@@ -126,7 +127,7 @@ TEST_F(BackendTest, FeedbackLoops) {
         TrianglePrimitive const triangle(getDriverApi());
 
         // Create a texture.
-        auto usage = TextureUsage::COLOR_ATTACHMENT | TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE;
+        auto usage = TextureUsage::COLOR_ATTACHMENT | TextureUsage::SAMPLEABLE;
         Handle<HwTexture> const texture = cleanup.add(api.createTexture(
             SamplerType::SAMPLER_2D, kNumLevels, kTexFormat, 1, kTexWidth, kTexHeight, 1, usage));
 

filament/backend/test/test_LoadImage.cpp

@@ -48,7 +48,7 @@ layout(location = 0) out vec4 fragColor;
 
 // Filament's Vulkan backend requires a descriptor set index of 1 for all samplers.
 // This parameter is ignored for other backends.
-layout(binding = 0, set = 0) uniform {samplerType} test_tex;
+layout(location = 0, set = 0) uniform {samplerType} test_tex;
 
 void main() {
     vec2 fbsize = vec2({texSize});
@@ -66,7 +66,7 @@ std::string fragmentUpdateImage3DTemplate (R"(#version 450 core
 layout(location = 0) out vec4 fragColor;
 
 // Filament's Vulkan backend requires a descriptor set index of 1 for all samplers.
-layout(binding = 0, set = 0) uniform {samplerType} test_tex;
+layout(location = 0, set = 0) uniform {samplerType} test_tex;
 
 float getLayer(in sampler3D s) { return 2.5f / 4.0f; }
 float getLayer(in sampler2DArray s) { return 2.0f; }
@@ -87,7 +87,7 @@ std::string fragmentUpdateImageMip (R"(#version 450 core
 layout(location = 0) out vec4 fragColor;
 
 // Filament's Vulkan backend requires a descriptor set index of 1 for all samplers.
-layout(binding = 0, set = 0) uniform sampler2D test_tex;
+layout(location = 0, set = 0) uniform sampler2D test_tex;
 
 void main() {
     vec2 fbsize = vec2({texSize});
@@ -231,7 +231,6 @@ public:
 };
 
 TEST_F(LoadImageTest, UpdateImage2D) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     FAIL_IF(Backend::VULKAN, "Multiple test cases crash, see b/417481434");
 
     // All of these test cases should result in the same rendered image, and thus the same hash.
@@ -272,18 +271,15 @@ TEST_F(LoadImageTest, UpdateImage2D) {
 
     // Test format conversion.
     // TODO: Vulkan crashes with `Texture at colorAttachment[0] has usage (0x01) which doesn't specify MTLTextureUsageRenderTarget (0x04)'
-    // TODO: WebGPU Crashes with "destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment"
     testCases.emplace_back("RGBA FLOAT to RGBA16F", PixelDataFormat::RGBA, PixelDataType::FLOAT, TextureFormat::RGBA16F);
 
     // Test texture formats not all backends support natively.
     // TODO: Vulkan crashes with "VK_FORMAT_R32G32B32_SFLOAT is not supported"
-    // TODO: WebGPU Crashes with "destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment"
     testCases.emplace_back("RGB FLOAT to RGB32F", PixelDataFormat::RGB, PixelDataType::FLOAT, TextureFormat::RGB32F);
     testCases.emplace_back("RGB FLOAT to RGB16F", PixelDataFormat::RGB, PixelDataType::FLOAT, TextureFormat::RGB16F);
 
     // Test packed format uploads.
     // TODO: Vulkan crashes with "Texture at colorAttachment[0] has usage (0x01) which doesn't specify MTLTextureUsageRenderTarget (0x04)"
-    // TODO: WebGPU Crashes with "destination texture usage doesn't have wgpu::TextureUsage::RenderAttachment"
     testCases.emplace_back("RGBA UINT_2_10_10_10_REV to RGB10_A2", PixelDataFormat::RGBA, PixelDataType::UINT_2_10_10_10_REV, TextureFormat::RGB10_A2);
     testCases.emplace_back("RGB UINT_10F_11F_11F_REV to R11F_G11F_B10F", PixelDataFormat::RGB, PixelDataType::UINT_10F_11F_11F_REV, TextureFormat::R11F_G11F_B10F);
     testCases.emplace_back("RGB HALF to R11F_G11F_B10F", PixelDataFormat::RGB, PixelDataType::HALF, TextureFormat::R11F_G11F_B10F);
@@ -291,12 +287,9 @@ TEST_F(LoadImageTest, UpdateImage2D) {
     // Test integer format uploads.
     // TODO: These cases fail on OpenGL and Vulkan.
     // TODO: These cases now also fail on Metal, but at some point previously worked.
-    // TODO: These cases fail for WebGPU. However, leaving them causes
-    //       Tint Reader Error: error: sampled image must have float component type
-    //       Beginning SpirV-output dump with ret 0
-//     testCases.emplace_back("RGB_INTEGER UBYTE to RGB8UI", PixelDataFormat::RGB_INTEGER, PixelDataType::UBYTE, TextureFormat::RGB8UI);
-//     testCases.emplace_back("RGB_INTEGER USHORT to RGB16UI", PixelDataFormat::RGB_INTEGER, PixelDataType::USHORT, TextureFormat::RGB16UI);
-//     testCases.emplace_back("RGB_INTEGER INT to RGB32I", PixelDataFormat::RGB_INTEGER, PixelDataType::INT, TextureFormat::RGB32I);
+     testCases.emplace_back("RGB_INTEGER UBYTE to RGB8UI", PixelDataFormat::RGB_INTEGER, PixelDataType::UBYTE, TextureFormat::RGB8UI);
+     testCases.emplace_back("RGB_INTEGER USHORT to RGB16UI", PixelDataFormat::RGB_INTEGER, PixelDataType::USHORT, TextureFormat::RGB16UI);
+     testCases.emplace_back("RGB_INTEGER INT to RGB32I", PixelDataFormat::RGB_INTEGER, PixelDataType::INT, TextureFormat::RGB32I);
 
     // Test uploads with buffer padding.
     // TODO: Vulkan crashes with "Assertion failed: (offset + size <= allocationSize)"
@@ -570,7 +563,7 @@ TEST_F(LoadImageTest, UpdateImage3D) {
     PixelDataType pixelType = PixelDataType::FLOAT;
     TextureFormat textureFormat = TextureFormat::RGBA16F;
     SamplerType samplerType = SamplerType::SAMPLER_2D_ARRAY;
-    TextureUsage usage = TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE | TextureUsage::COLOR_ATTACHMENT;
+    TextureUsage usage = TextureUsage::SAMPLEABLE | TextureUsage::UPLOADABLE;
 
     // Create a platform-specific SwapChain and make it current.
     auto swapChain = cleanup.add(createSwapChain());

filament/backend/test/test_ReadPixels.cpp

@@ -80,7 +80,6 @@ public:
 };
 
 TEST_F(ReadPixelsTest, ReadPixels) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     NONFATAL_FAIL_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Two cases fail, see b/417255941 and b/417255943");
     // These test scenarios use a known hash of the result pixel buffer to decide pass / fail,

filament/backend/test/test_RenderExternalImage.cpp

@@ -46,9 +46,8 @@ Shader createShader(DriverApi& api, Cleanup& cleanup, Backend backend) {
 
 // Rendering an external image without setting any data should not crash.
 TEST_F(BackendTest, RenderExternalImageWithoutSet) {
-    SKIP_IF(Backend::METAL, "External images aren't supported in Metal");
-    SKIP_IF(Backend::VULKAN, "External images aren't supported in Vulkan");
-    SKIP_IF(Backend::WEBGPU, "External images aren't supported in WebGPU");
+    SKIP_IF(Backend::METAL, "External images aren't supported in metal");
+    SKIP_IF(Backend::VULKAN, "External images aren't supported in vulkan");
     auto& api = getDriverApi();
     Cleanup cleanup(api);
 
@@ -110,9 +109,8 @@ TEST_F(BackendTest, RenderExternalImageWithoutSet) {
 }
 
 TEST_F(BackendTest, RenderExternalImage) {
-    SKIP_IF(Backend::METAL, "External images aren't supported in Metal");
-    SKIP_IF(Backend::VULKAN, "External images aren't supported in Vulkan");
-    SKIP_IF(Backend::WEBGPU, "External images aren't supported in WebGPU");
+    SKIP_IF(Backend::METAL, "External images aren't supported in metal");
+    SKIP_IF(Backend::VULKAN, "External images aren't supported in vulkan");
     auto& api = getDriverApi();
     Cleanup cleanup(api);
 

filament/backend/test/test_Scissor.cpp

@@ -31,7 +31,6 @@ using namespace filament;
 using namespace filament::backend;
 
 TEST_F(BackendTest, ScissorViewportRegion) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     NONFATAL_FAIL_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Affected area in wrong corner, see b/417229118");
     auto& api = getDriverApi();

filament/backend/test/test_StencilBuffer.cpp

@@ -126,7 +126,6 @@ public:
 };
 
 TEST_F(BasicStencilBufferTest, StencilBuffer) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     SKIP_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Stencil not supported, see b/417230776");
     auto& api = getDriverApi();
@@ -153,7 +152,6 @@ TEST_F(BasicStencilBufferTest, StencilBuffer) {
 }
 
 TEST_F(BasicStencilBufferTest, DepthAndStencilBuffer) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     SKIP_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Stencil not supported, see b/417230776");
     auto& api = getDriverApi();
@@ -180,7 +178,6 @@ TEST_F(BasicStencilBufferTest, DepthAndStencilBuffer) {
 }
 
 TEST_F(BasicStencilBufferTest, StencilBufferMSAA) {
-    SKIP_IF(Backend::WEBGPU, "test cases fail in WebGPU, see b/424157731");
     SKIP_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::VULKAN),
             "Stencil not supported, see b/417230776");
     SKIP_IF(SkipEnvironment(OperatingSystem::APPLE, Backend::OPENGL), "Stencil isn't applied");