[FFmpeg-devel] [PATCH 1/7] avcodec/h263dec: Remove redundant code to set cur_pic_ptr
Andreas Rheinhardt
andreas.rheinhardt at outlook.com
Thu Aug 18 01:05:44 EEST 2022
Michael Niedermayer:
> On Mon, Aug 15, 2022 at 01:49:24PM +0200, Andreas Rheinhardt wrote:
>> It is done later in ff_mpv_frame_start() (and nobody uses
>> current_picture_ptr between setting it in ff_mpv_frame_start()).
>>
>> (The reason the vsynth*-h263-obmc code changes is because
>> the call to ff_find_unused_picture() now happens after the older
>> pictures have been unreferenced in ff_mpv_frame_start(),
>> so that their slots in the picture array can be immediately
>> reused; the obmc code is somehow buggy and changes its output
>> depending on the earlier contents of the motion_val buffer.)
>>
>> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at outlook.com>
>> ---
>
>> I'd like to take this opportunity to once again ask anyone familiar
>> with H.263 to take a look at this OBMC issue.
>
> Iam too busy ATM :( i might look at some point but its not very high
> on my todo as it works :) security & release is more important
>
If you (and no one else) don't object, I'll apply this tomorrow.
> but i can say this breaks fate as it is:
>
> --- ./tests/ref/vsynth/vsynth1-h263-obmc 2022-08-16 00:19:00.345967181 +0200
> +++ tests/data/fate/vsynth1-h263-obmc 2022-08-16 00:19:05.262017999 +0200
> @@ -1,4 +1,4 @@
> 7dec64380f375e5118b66f3baaaa1e24 *tests/data/fate/vsynth1-h263-obmc.avi
> 657320 tests/data/fate/vsynth1-h263-obmc.avi
> -f5048b5f0c98833a1d11f8034fb1827f *tests/data/fate/vsynth1-h263-obmc.out.rawvideo
> -stddev: 8.12 PSNR: 29.93 MAXDIFF: 113 bytes: 7603200/ 7603200
> +2a69f6b37378aa34418dfd04ec98c1c8 *tests/data/fate/vsynth1-h263-obmc.out.rawvideo
> +stddev: 8.38 PSNR: 29.66 MAXDIFF: 116 bytes: 7603200/ 7603200
> Test vsynth1-h263-obmc failed. Look at tests/data/fate/vsynth1-h263-obmc.err for details.
> tests/Makefile:304: recipe for target 'fate-vsynth1-h263-obmc' failed
> make: *** [fate-vsynth1-h263-obmc] Error 1
>
> [...]
>
More information about the ffmpeg-devel
mailing list