* Document better BRDFs
- Document the loss of energy in the specular BRDF
- Start documenting the gain of energy in the diffuse BRDF
* Explain energy compensation with second scaled lobe
* Add multiscattering pre-integration
* better froxel records compression
We used to compress froxel records by reusing the
previous record on the left if it was identical,
now we additionally check the record above if
that fails.
In practice this saves 10% to 30% space in the
froxel record buffer, which is a scarce resource
for us.
e.g. if a light occupied 3x3 froxels in a given
z-slice, without compression 9 records would be
used, with left-only compression, only 3,
and with left+above compression, only 1.
* use size_t for in and out parameters
this helps the compiler in many places, not
having to cast from 64 to 32 bits.
* implement bitset’s any() and all() operations with NEON
* empty froxels are a common case, handle it first
This changes the layout of the froxels data
structure in the GPU so that the major axis
is the Z axis instead of X. The idea is that
within a given screen-space tile, we have a high
probability to access froxels at the same x/y
but different z.
This is turned off right now, as it requires
a change that breaks existing compiled materials.
apparently off-by-one’s hard. The first row of
our *source* froxel data must be skipped because
it contains a bit mask representing wether a light
is point or spot.
* Enhance CMake compiler check
This commit adds a check for compilers that don't meet the project's
minimum requirements. It will catch Clang versions that are too low as
well as any non-Clang compilers.
* Change SEND_ERROR to FATAL_ERROR
* Update SPIRV libraries
- Update `spirv-cross`
- Update `spirv-tools`
- Update `glslang`
`spirv-tools` contains a fix for an issue we reported (unnecessary promotions to highp in certain expressions).
* Add SPIRV-Tools dependency to matinfo
* Add material_testbed application
This application can be used to automatically vary material parameters over multiple frames and capture each frame as a PNG file. This makes it easier to test/compare/etc. It's also useful to generate documentation examples.
* Rename material_testbed to frame_generator
* JobSystem now automatically free Jobs
Until now Job allocation used a linear allocator
strategy which required to “reset” the JobSystem
periodically — typically once per frame in
filament.
This is no longer required. We use a pool allocator
now, which doesn’t add much overhead. It does
use a spin-lock for thread-safety though, since
we assume very little contention, this shouldn’t
be a problem.
* Thread Safe Object Pool Allocator
A lock-less, thread-safe object pool allocator,
now used for storing JobSystem’s jobs allocations.
This gets rid of the spin-lock introduced in the
previous cl.
We recently changed the default constructor of vector to not initialize the vector to 0. This caused subtle issues in cmgen which was creating arrays of vectors. This change simply forces 0 initialization of those arrays.
fix#87
We just skip samples that have a dot(N,L)<0 which
means we’re “loosing” some samples. However,
this happens higher levels of the mipmap chain,
and since we’re increasing significantly the
number of samples as the level increases, this
doesn’t end-up being a problem.
- handle errors during context creation
- handle errors when setting the swap chain
- don’t leak the context and dummy window
- don’t return dummy fences that don’t even work
Romain Guy
Romain Guy
Mathias Agopian
Philip Rideout
Adrian Perez
Mathias Agopian
Morten Olsen Lysgaard
Cory Bloor
shartte
Benjamin Doherty