[FFmpeg-devel] Official status of FFv1?

Jason Garrett-Glaser darkshikari
Mon Mar 22 22:00:36 CET 2010


On Mon, Mar 22, 2010 at 1:44 PM, Lars T?uber <lars.taeuber at gmx.net> wrote:
> Hi there!
>
> On Mon, 22 Mar 2010 13:11:24 -0700 Jason Garrett-Glaser <darkshikari at gmail.com> wrote:
>> Here's some benches for you, Baptiste. ?Source: standard foreman CIF.
>>
>> $ ffmpeg -i videos/foreman_cif.y4m -vcodec ffv1 -coder 1 output.avi
>> frame= ?300 fps= 48 q=0.0 Lsize= ? 19017kB time=10.01 bitrate=15563.4kbits/s
>> ----------+ 1 Jason None 19473662 Mar 22 13:07 output.avi
>>
>> $ ffmpeg -i videos/foreman_cif.y4m -vcodec ffv1 -coder 0 output.avi
>> frame= ?300 fps= 98 q=0.0 Lsize= ? 19270kB time=10.01 bitrate=15770.5kbits/s
>> ----------+ 1 Jason None 19732838 Mar 22 13:10 output.avi
>>
>> $ ./x264 videos/foreman_cif.y4m --preset veryfast --qp 0 -o test.h264
>> --threads 1
>> encoded 300 frames, 75.17 fps, 13583.06 kb/s
>> ----------+ 1 Jason None 16995802 Mar 22 13:07 test.h264
>>
>> $ ./x264 videos/foreman_cif.y4m --preset veryfast --qp 0 -o test.h264
>> --threads 1 --no-cabac;ls -l test.h264
>> encoded 300 frames, 161.38 fps, 15550.08 kb/s
>> ----------+ 1 Jason None 19457036 Mar 22 13:09 test.h264
>
> So its time to put P-frame support for FFV1 on the GSoC list, isn't it?
> Maybe the contribution of suppport for 10bit or 16bit (or weven floating point) colorspaces to ffv1 could be the qualification task?
>
> No, I am not smart enough to do this.

We (me and Loren) have a rough design for an FFV2 which would be
slightly worse than FFV1 intra-wise but add inter compression and be
faster than HuffYUV.  It would surely outperform x264.

If anyone is interested, I can lay out a full design document from my memory.

Dark Shikari



More information about the ffmpeg-devel mailing list