Scarfski/basis_universal
1
0
Fork 0
mirror of https://github.com/BinomialLLC/basis_universal.git synced 2026-08-16 08:09:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Rich Geldreich
2019-05-29 12:48:42 -07:00
committed by GitHub
parent 3050ca4a70
commit a2144d7ff4
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
README.md
View File

@@ -130,6 +130,14 @@ The "WebGL" directory contains two very simple WebGL demos that use the transcod
![Screenshot of 'texture' example running in a browser.](webgl/texture/preview.png)
![Screenshot of 'gltf' example running in a browser.](webgl/gltf/preview.png)
### .basis baseline file format forwards and backwards compatibility
Short form: Old transcoders will always be able to read baseline compressed texture data written by new encoders. And new transcoders will always still be able to read old .basis files generated by old encoders.
We have no intentions of changing the baseline file format (i.e. the file format written by this encoder), apart from critical bug fixes. We will be extending the format in the future to add features to the system, but old transcoders will still be able to transcode baseline texture data from .basis files written using new transcoders. .basis files are intended for very wide distribution of compressed texture data, like .jpeg or .png, so backwards and forwards compatibility is critical.
So if you compile and ship the transcoder into an application today, we are making a guarantee to you that this transcoder will stay relevant/usable even as we enhance the system. To do this, the file format has the ability to be extended.
### Transcoder details
The transcoder unpacks .basis files to various GPU texture formats, almost always without needing to decompress entire images at the pixel level (i.e. it only deals with arrays of blocks). The one exception is PVRTC1, where the transcoder needs to recompute the per-pixel selector ("modulation") values, but it does so using simple scalar operations.