[FFmpeg-devel] FATE tests: FFV1
Peter B.
pb at das-werkstatt.com
Sun Nov 3 21:53:38 CET 2013
On 11/03/2013 09:37 PM, Carl Eugen Hoyos wrote:
> 2) I smell surprise in the paragraph you wrote: Please
> feel free to test a (not too small) number of fixed
> FFmpeg tickets with avconv, you will find that there are
> several hundred regressions with avconv, don't you agree
> that it would have been more of a surprise if there were
> none in the FFV1 codec?
I wasn't surprised. Unfortunately.
I do see that ffmpeg's merging changes/improvements done in avconv, but
I'm really sad to see that it doesn't happen the other way around. Well...
>> Almost all files are now generated, so fate-suite impact is very little,
>> but this means that the tests add computation time, each time FATE is
>> ran completely.
> This is acceptable as long as only a few frames are encoded
> (as many as necessary).
I'm encoding vsynth1.yuv as-is.
For my own files for fuzz-testing, I'm only using 4 clean, generated
frames - so they compress good.
>> Furthermore, I saw that encoding tests (e.g. in "vcodecs.mak") for
>> lossless tests would actually require to have its own raw source file
>> for every tested parameter, like pix_fmt or resolution.
>> Resizing on the fly, or forcing the source to a different pix_fmt
>> results in "MAXDIFF > 0".
> I may misunderstand but imo this is just a question of
> the options used.
Could you point me at any documentation about these possible options to use?
> Of course creating artificial source files in different
> pix_fmts would be welcome!
I think I've already spotted where vsynth[12].yuv are generated...
Will take a look.
I guess that other FATE tests might also be happy to have common files
to refer to?
(e.g. Improving other lossless tests would then very easy)
> Please note that all these questions are development-related
> (particularly as you intend to send patches), so the
> discussion should better take place at ffmpeg-devel (which
> by the way is an open mailing list, suited for open source
> development...)
Moving it there.
Regards,
Pb
More information about the ffmpeg-devel
mailing list