[FFmpeg-devel] [PATCH] avcodec: add Actimagine VX video decoder

Florian Nouwt fnouwt2 at gmail.com
Tue Mar 16 14:23:41 EET 2021


I managed to utilize the h264 idct and h264 prediction code now. The
only prediction code I kept is for the 4 modes that work on any
possible block size (horizontal, vertical, dc and plane). I also moved
the actual vlc tables and their initialization into h264_cavlc_data.c
and checked that make fate was successful. I will await the discussion
on the color format before submitting the new patch. I have attached 3
images that show an original video frame I encoded to the vx format,
the version decoded with the windows codec and the version decoded by
ffmpeg utilizing the regular yuv formulas. Before I implemented this
format in ffmpeg I had written a decoder in C# which utilized the
color conversion formulas I discussed before, and the output of that
program was exactly equal to the frame as decoded by the windows
codec. So I think that should be enough evidence that those formulas
are right.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sm3dw_win_codec.png
Type: image/png
Size: 56295 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20210316/758d20ed/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sm3dw_original.png
Type: image/png
Size: 71030 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20210316/758d20ed/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sm3dw_ffmpeg.png
Type: image/png
Size: 77714 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20210316/758d20ed/attachment-0002.png>


More information about the ffmpeg-devel mailing list