[FFmpeg-devel] [PATCH v1] avcodec/av1_parser: ensure only one show frame packed data parsered

[Wang, Fei W]

On Fri, 2021-12-03 at 09:36 -0300, James Almer wrote:
> On Fri, Dec 3, 2021 at 5:12 AM Fei Wang <fei.w.wang at intel.com> wrote:
> 
> > Split packed data when it contains multiple show frames in some
> > non-standard bitstream. This can benefit downstream decoder which
> > can
> > decode continuously instead of interrupt with unexpected error.
> > 
> > Signed-off-by: Fei Wang <fei.w.wang at intel.com>
> > ---
> >  libavcodec/av1_parser.c | 19 +++++++++++++++----
> >  1 file changed, 15 insertions(+), 4 deletions(-)
> > 
> > diff --git a/libavcodec/av1_parser.c b/libavcodec/av1_parser.c
> > index d2dfdb3580..d6f5cace4b 100644
> > --- a/libavcodec/av1_parser.c
> > +++ b/libavcodec/av1_parser.c
> > @@ -59,11 +59,10 @@ static int
> > av1_parser_parse(AVCodecParserContext *ctx,
> >      const CodedBitstreamAV1Context *av1 = s->cbc->priv_data;
> >      const AV1RawSequenceHeader *seq;
> >      const AV1RawColorConfig *color;
> > +    int pic_found = 0;
> > +    int next = 0;
> >      int ret;
> > 
> > -    *out_data = data;
> > -    *out_size = size;
> > -
> >      ctx->key_frame         = -1;
> >      ctx->pict_type         = AV_PICTURE_TYPE_NONE;
> >      ctx->picture_structure = AV_PICTURE_STRUCTURE_UNKNOWN;
> > @@ -100,6 +99,8 @@ static int av1_parser_parse(AVCodecParserContext
> > *ctx,
> >          const AV1RawOBU *obu = unit->content;
> >          const AV1RawFrameHeader *frame;
> > 
> > +        next += unit->data_size;
> > +
> >          if (unit->type == AV1_OBU_FRAME)
> >              frame = &obu->obu.frame.header;
> >          else if (unit->type == AV1_OBU_FRAME_HEADER)
> > @@ -113,6 +114,12 @@ static int
> > av1_parser_parse(AVCodecParserContext *ctx,
> >          if (!frame->show_frame && !frame->show_existing_frame)
> >              continue;
> > 
> > +        /* split data if it contains multi show frames */
> > +        if (pic_found) {
> > +            next -= unit->data_size;
> > +            break;
> > +        }
> > +
> >          ctx->width  = frame->frame_width_minus_1 + 1;
> >          ctx->height = frame->frame_height_minus_1 + 1;
> > 
> > @@ -131,8 +138,12 @@ static int
> > av1_parser_parse(AVCodecParserContext *ctx,
> >              break;
> >          }
> >          ctx->picture_structure = AV_PICTURE_STRUCTURE_FRAME;
> > +        pic_found = 1;
> >      }
> > 
> > +    *out_size = next;
> > +    *out_data = data;
> > +
> >      switch (av1->bit_depth) {
> >      case 8:
> >          ctx->format = color->mono_chrome ? AV_PIX_FMT_GRAY8
> > @@ -171,7 +182,7 @@ end:
> > 
> >      s->cbc->log_ctx = NULL;
> > 
> > -    return size;
> > +    return next;
> >  }
> > 
> >  static const CodedBitstreamUnitType decompose_unit_types[] = {
> > --
> > 2.25.1
> > 
> > 
> No, the parser must not split the packets. We used to do this with
> the VP9
> parser and the result was the creation of broken output files.
Any more information on failed case for vp9? I tested this patch by
decoding several av1 bitstream, all looks good.

And I checked wm4's change on vp9/vp9_parser, the commit only shows it
benefits on multi-thread decoding. So I guess if vp9 failed case is
related to multi-thread decodeing? If yes, then insert frame_split bsf
in av1 decoder will be a better choice. Even av1 decoder doesn't
support multi-threading decode now, it can avoid issue if it is
supported someday.

> 
> What you could do instead is autoinsert the frame_split BSF in the
> decoder,
> same as we're doing for VP9. It should have the same effect.
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> 
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".