Follow up from #1438: we could decode webp to yuv if we encoded it from yuv data (the code currently only supports rgb input) and told the client that it can safely use yuv without losing accuracy.
This should give us a significant performance and compression boost (use with quality settings lower than ~50%?), maybe even enough to supplant jpeg for all settings?
We would need a csc step... and not one managed by the video pipeline setup code. Maybe just hardcode a dependency on libyuv for this feature?
Done in r18334 (see long commit message) since it was easy enough to add using webp's WebPPictureARGBToYUVA
for colorspace conversion.
It would be nice to also paint YUVA: #1765.
Caused a regression: #1769
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1764