On Mon, 6 May 2019 03:33:58 +0000 Manoj Bonda <mbo...@nvidia.com> wrote:
> Thanks Philip for pushing the changes, sorry for the churn. > > I am not familiar with the MPV code much, but from a quick check by > enabling the direct mode for 444 surfaces by modifying the condition > > p->direct_mode = mapper->dst_params.hw_subfmt == IMGFMT_NV12 || > mapper->dst_params.hw_subfmt == IMGFMT_420P || > to > p->direct_mode = mapper->dst_params.hw_subfmt == IMGFMT_NV12 || > mapper->dst_params.hw_subfmt == IMGFMT_420P || > mapper->dst_params.hw_subfmt == IMGFMT_444P; > > the playback seems to be fine. But I see this might be since we are > downscaling the chroma texture in function mapper_map ? > > I don’t see semi-planar formats for 444 in MPV code. > VDPAU OpenGL interop supports only semiplanar surfaces as per the > spec > (https://www.khronos.org/registry/OpenGL/extensions/NV/NV_vdpau_interop2.txt) > I see the proper way to enable direct mode in MPV is to add support > for 444 semiplanar format. Please correct me if I am wrong. I'm surprised it doesn't look more mangled than it does, because the code assumes the format is always NV12 right now. That means it'll take the top left quarter of the chroma plane and use that. Maybe my sample hides the effect. Anyway, yes, it means we need to add an NV24 pix fmt to ffmpeg and then use it in mpv. > Sorry, but I won’t have bandwidth to take up this MPV task now. I > will try to get back to it once I find some time. I'll start looking at it. Thanks for investigating. --phil _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
