[FFmpeg-devel] [PATCH] Add additional FFV1 fate tests

[Michael Niedermayer]

On Sun, Nov 10, 2013 at 11:16:48PM +0100, Peter B. wrote:
> On 11/10/2013 05:41 PM, Michael Niedermayer wrote:
> > On Sun, Nov 10, 2013 at 04:52:28PM +0100, Peter B. wrote:
> >> How is the default procedure for patch-improvements?
> >> Should I post a patch for my patch - or rather one patch that applies as
> >> one to git head?
> > one patch on top of git head in this case
> 
> Clear.
> I've attached the new patch.
> 
> Changes (compared to the previous version) include:
> - Setting the word size for tiny_psnr to 8bits/16bits accordingly
> - ${} replaced with $()
> - Aligned "\" for multi-line statements
> 
> 
> I'm still uncomfortable that the MAXDIFF ain't 0, which is caused by
> re-using vsynth1 in different pix_fmts. A value >0 seems a bit confusing
> for a lossless codec :)
> However, as I understood it, the values calculated by tiny_psnr will be
> constant in case of FFV1, so they'll match the ref-files and therefore
> should be fine for testing. I'm planning to improve this situation by
> generating dedicated 4-frame-testvideos for multiple pix_fmts and
> resolutions on-the-fly (like vsynth) in the future.

running make -j12 `make fate-list | grep ffv1`
i get

TEST    ffv1-dec-v3-yuv422p_pass1
TEST    ffv1-enc-v3-yuv422p_pass2
TEST    ffv1-dec-v3-yuv444p
TEST    ffv1-dec-v3-yuv444p10
TEST    ffv1-dec-v3-yuv444p16
TEST    ffv1-dec-v3-yuv444p9
TEST    ffv1-dec-v3-yuva420p10
TEST    ffv1-dec-v3-yuva420p16
TEST    ffv1-dec-v3-yuva420p9
TEST    ffv1-dec-v3-yuva422p10
make: `fate-ffv1-enc-v3-yuv422p_pass1' is up to date.
make: `fate-ffv1-enc-v3-yuv444p16' is up to date.
make: `fate-ffv1-enc-v3-yuv444p9' is up to date.
make: `fate-ffv1-enc-v3-yuva420p10' is up to date.
make: `fate-ffv1-enc-v3-yuva420p16' is up to date.
make: `fate-ffv1-enc-v3-yuva420p9' is up to date.
make: `fate-ffv1-enc-v3-yuva422p10' is up to date.
make: `fate-ffv1-enc-v3-yuva422p9' is up to date.
TEST    ffv1-dec-v3-yuva422p9
make: `fate-ffv1-enc-v3-yuva422p16' is up to date.
TEST    ffv1-dec-v3-yuva422p16
make: `fate-ffv1-enc-v3-yuva444p10' is up to date.
TEST    ffv1-dec-v3-yuva444p10
make: `fate-ffv1-enc-v3-yuv422p_pass2' is up to date.
TEST    ffv1-dec-v3-yuv422p_pass2
make: `fate-ffv1-enc-v3-yuva444p9' is up to date.
TEST    ffv1-dec-v3-yuva444p9
make: `fate-ffv1-enc-v3-yuva444p16' is up to date.
TEST    ffv1-dec-v3-yuva444p16
make: `fate-vsynth2-ffv1' is up to date.
TEST    seek-vsynth2-ffv1

a simpler testcase is
make fate-ffv1-dec-v1-bff fate-ffv1-enc-v1-bff -j12
TEST    ffv1-enc-v1-bff
make: `fate-ffv1-enc-v1-bff' is up to date.
TEST    ffv1-dec-v1-bff

about the fuzzed samples, cant tools/trasher be used to generate
them from some source ffv1 ?
or is there something special about them that makes this
disadvantaneous ?

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Old school: Use the lowest level language in which you can solve the problem
            conveniently.
New school: Use the highest level language in which the latest supercomputer
            can solve the problem without the user falling asleep waiting.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20131111/ac4ef26f/attachment.asc>