>> Temporal AQ is implemented in CUDA, not in silicon. So far does the API (Video SDK 7.0.1) accept the flag without giving any errors. The output however remains identical currently (tested on a GTX 960). You'll have to ask Nvidia why they haven't implemented it yet (i'ts probably too soon). I am expecting it to become available with a future driver (or SDK or CUDA). It's already available for NVENC when encoding H.264/AVC and it's not only a method used in H.264 encoding, but one which is used in H.265 encoding, too (for example does x265 call auto variance '--aq-mode 2').
I am from Nvidia Video Team, so can give a good answer. Currently it has been implemented for H264 only. HEVC TAQ may be released in a future SDK release. All queries for NVENC/CUVID can also be directed to me. Thanks, Yogender _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel