[FFmpeg-devel] [PATCH] Enable PAFF decoding

[Andreas Öman]

Hi,

Neil Brown wrote:

> 2/ The frames are not 100% identical to those generated by the
>    reference decoder.
> 
>  I use
>     ffmpeg -i test.mts -t 0.2 /tmp/ff%d.ppm
>  and
>     ffmpeg -s 1440x1080 -i test.yuv -t 0.2 /tmp/ref%d.ppm
> 
>  to collect the frames, find two that look the same (i.e. compensate
>  for the ordering problems), load them as two different layers into
>  GIMP, and use the "Difference" Mode to find the difference.
> 
>  The result looks black, but when you look at the histogram or adjust
>  the thresholds, there are differences, mostly 1 or 2, but as much as
>  10.
>  I won't be losing sleep over this, but I thought you should know.

Seems to be the case with non-PAFF content as well.
Not all content though.
The run below compares raw output from ffmpeg and ldecod using file
called 'ffh264_issue3.h264' which has been posted earlier here.
I think it's a bluray rip or something.

The output is from a hack i've done that compares raw yuv.
The tool just dumps mismatching blocks from two streams side by side
and puts <> around pixels that mismatches.

Only the chroma planes are mismatching.
In total 28042 macroblocks are differs over 280 frames.
I have not done any futher analysis yet, so i don't even know
if it related to motion comp or not.

[...]
Comparing frame 221
==========================================================================
MB ( 58,48 ) 5818 Cr mismatches
Chroma-r
  94  95  96  97  97  97  97  97  |  94  95  96  97  97  97  97  97
  9a  9b  9c >9d<>9f< 9f  9f  9f  |  9a  9b  9c >9e<>9e< 9f  9f  9f
  9f  a0  a1  a2  a2  a2  a2  a2  |  9f  a0  a1  a2  a2  a2  a2  a2
  a1  a2  a3 >a4< a1  a1  a1  a1  |  a1  a2  a3 >a3< a1  a1  a1  a1
  9e  9e  9e >9e< a0  a0  a0  a0  |  9e  9e  9e >9f< a0  a0  a0  a0
  9e  9e  9e >9e<>a0< a0  a0  a0  |  9e  9e  9e >9f<>9f< a0  a0  a0
  9b  9b  9b >9b<>9d< 9d  9d  9d  |  9b  9b  9b >9c<>9c< 9d  9d  9d
  98  98  98 >98<>9a< 9a  9a  9a  |  98  98  98 >99<>99< 9a  9a  9a

Some blocks are more damaged:
==========================================================================
MB ( 86,49 ) 5966 Cr mismatches
Chroma-r
 >75<>76<>76<>76<>76<>76<>76<>76< | >77<>78<>78<>78<>78<>78<>78<>78<
 >75<>75<>75<>75<>75<>76<>76<>76< | >76<>77<>77<>77<>77<>78<>78<>78<
 >74<>75<>75<>75<>75<>75<>75<>75< | >75<>76<>76<>76<>76<>76<>76<>77<
 >74<>75<>75<>75<>75<>74<>74<>74< | >75<>76<>76<>76<>76<>75<>75<>75<
 >74<>75<>75<>75<>75<>74<>74<>74< | >75<>76<>76<>76<>76<>75<>75<>76<
 >73<>74<>74<>74<>75<>74<>74<>74< | >75<>75<>75<>75<>76<>75<>75<>75<
 >73<>74<>74<>74<>74<>74<>74<>74< | >75<>75<>75<>75<>75<>75<>75<>75<
 >73<>74<>74<>74<>74<>74<>74<>74< | >74<>75<>75<>75<>75<>75<>75<>75<