20:11 <jenneron[m]> digetx: hi, could you explain me some things please, are weighted prediction and cabac only user-space part or both kernel and user-space?

20:11 <jenneron[m]> there is translator v4l -> va-api - is it be usable for us?

20:11 <jenneron[m]> https://github.com/bootlin/libva-v4l2-request

20:11 <jenneron[m]> will*

20:13 <digetx> weighted prediction and cabac are both

20:13 <digetx> libva-v4l2-request is deprecated

20:14 <jenneron[m]> there is a PR with 5.14+ kernel support and h264

20:14 <jenneron[m]> do you mean there is a better vaapi implementation with v4l?

20:17 <digetx> there is no other implementation to my knowledge

20:18 <digetx> you could translate vdpau to vaapi, but why would you want to do it?

20:21 <digetx> vaapi requires gl and using gl for video output isn't great for mobile devices because of the overhead

20:26 <digetx> but if you would want to accel decoding in a web browser, then there is no better solution today than to use gl

20:27 <digetx> many years ago there was a dream that wayland will fix it all, but that's still a dream :)

20:29 <digetx> it's a difficult problem to develop api for efficient displaying of multimedia and then make it widespread

20:32 <jenneron[m]> iirc wayland doesn't support vdpau, so in long term we would need va-api