[FFmpeg-devel] [PATCH 1/2] lavc/cbs_h265: Add sps_multilayer_extension support

[Mark Thompson]

On 17/11/2023 13:03, Jun Zhao wrote:
> Add sps_multilayer_extensio support.
> 
> Signed-off-by: Jun Zhao <barryjzhao at tencent.com>
> ---
>   libavcodec/cbs_h265.h                 |  3 +++
>   libavcodec/cbs_h265_syntax_template.c | 12 +++++++++++-
>   2 files changed, 14 insertions(+), 1 deletion(-)
> 
> diff --git a/libavcodec/cbs_h265.h b/libavcodec/cbs_h265.h
> index 1b1195f198..15951269fd 100644
> --- a/libavcodec/cbs_h265.h
> +++ b/libavcodec/cbs_h265.h
> @@ -330,6 +330,9 @@ typedef struct H265RawSPS {
>       uint8_t persistent_rice_adaptation_enabled_flag;
>       uint8_t cabac_bypass_alignment_enabled_flag;
>   
> +    // Multilayer extension.
> +    uint8_t inter_view_mv_vert_constraint_flag;
> +
>       // Screen content coding extension.
>       uint8_t sps_curr_pic_ref_enabled_flag;
>       uint8_t palette_mode_enabled_flag;
> diff --git a/libavcodec/cbs_h265_syntax_template.c b/libavcodec/cbs_h265_syntax_template.c
> index 2d4b954718..1e3bc1acd8 100644
> --- a/libavcodec/cbs_h265_syntax_template.c
> +++ b/libavcodec/cbs_h265_syntax_template.c
> @@ -717,6 +717,16 @@ static int FUNC(sps_range_extension)(CodedBitstreamContext *ctx, RWContext *rw,
>       return 0;
>   }
>   
> +static int FUNC(sps_multilayer_extension)(CodedBitstreamContext *ctx, RWContext *rw,
> +                                          H265RawSPS *current)
> +{
> +    int err;
> +
> +    flag(inter_view_mv_vert_constraint_flag);
> +
> +    return 0;
> +}
> +
>   static int FUNC(sps_scc_extension)(CodedBitstreamContext *ctx, RWContext *rw,
>                                      H265RawSPS *current)
>   {
> @@ -952,7 +962,7 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
>       if (current->sps_range_extension_flag)
>           CHECK(FUNC(sps_range_extension)(ctx, rw, current));
>       if (current->sps_multilayer_extension_flag)
> -        return AVERROR_PATCHWELCOME;
> +        CHECK(FUNC(sps_multilayer_extension)(ctx, rw, current));
>       if (current->sps_3d_extension_flag)
>           return AVERROR_PATCHWELCOME;
>       if (current->sps_scc_extension_flag)

This doesn't seem like it it sufficient - when in multilayer the SPS format isn't the same (§F.7.3.2.2.1).

Alternatively: maybe this works for the first layer (haven't verified this but seems plausible), and therefore to work there only it would be sufficient to check if MultiLayerExtSpsFlag is true in an SPS and stop if it is?

(A sample suitable for using with fate would help.)

Thanks,

- Mark