[FFmpeg-devel] [PATCH] Enable PAFF decoding

[Jeff Downs]

On Wed, 10 Oct 2007, Neil Brown wrote:
[...]
> 3 issues:
> 
> 1/ The first few frames are in the wrong order.
> 
> The first 3 field-pairs in my files are:
>   I+P  (poc 0 and 1)
>   B+B  (poc -4 and -3)
>   B+B  (poc -2 and -1)
> 
> I'm getting the first pair out first, then the second decodes
> imperfectly.  I haven't figured out why yet.  The same thing doesn't
> happen when you cross an IDR frame, so there must be something
> "special" about the very start.

Do your files start with IDR?

 
> 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.

Me neither :)  Maybe someone else will, but as Andreas posted (thanks) it 
happens with non-PAFF so I'm not going to concentrate on this one.

 
> 3/ I'm getting the error message:
>   [h264 @ 0xa638a0]MBAFF + spatial direct mode is not implemented
> 
>   a lot. As the file doesn't use MBAFF (only PAFF) I suspect this is
>   wrong.
> 
>   Maybe this patch is correct?
> 
> Thanks,
> 
> NeilBrown
> 
> ---
>  libavcodec/h264.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> --- ffmpeg.orig/libavcodec/h264.c
> +++ ffmpeg/libavcodec/h264.c
> @@ -4396,7 +4396,7 @@ static int decode_slice_header(H264Conte
>      if(h->slice_type == P_TYPE || h->slice_type == SP_TYPE || h->slice_type == B_TYPE){
>          if(h->slice_type == B_TYPE){
>              h->direct_spatial_mv_pred= get_bits1(&s->gb);
> -            if(h->sps.mb_aff && h->direct_spatial_mv_pred)
> +            if(h->mb_aff_frame && h->direct_spatial_mv_pred)
>                  av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF + spatial direct mode is not implemented\n");
>          }
>          num_ref_idx_active_override_flag= get_bits1(&s->gb);


Looks right to me. This was actually in my orignal PAFF patch (I think 
taken from your priliminary PAFF patch of long ago) because I do get 
similar errors.  I pulled it out during the splitting process.

The problem is that the SPS has to include the MBAFF flag for any 
interlaced sequence (including PAFF). But for a PAFF frame, its value is 
meaningless; an encoder could freely set it either way. mb_aff_frame is 
the correct flag to test.

	-Jeff