[FFmpeg-devel] [PATCH 2/7] lavc/hevc_ps: Add sps parse support for HEVC SCC extension syntax

Linjie Fu linjie.justin.fu at gmail.com
Thu Jul 9 20:35:06 EEST 2020


On Fri, Jul 10, 2020 at 12:54 AM James Almer <jamrial at gmail.com> wrote:
>
> On 7/9/2020 1:51 PM, James Almer wrote:
> > On 7/9/2020 1:17 PM, Linjie Fu wrote:
> >> According to 7.3.2.2.3 in T-REC-H.265-201911.
> >>
> >> Signed-off-by: Linjie Fu <linjie.justin.fu at gmail.com>
> >> ---
> >>  libavcodec/hevc.h    |  3 +++
> >>  libavcodec/hevc_ps.c | 32 +++++++++++++++++++++++++++++---
> >>  libavcodec/hevc_ps.h | 15 +++++++++++++++
> >>  3 files changed, 47 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/libavcodec/hevc.h b/libavcodec/hevc.h
> >> index 1804755..6b454a7 100644
> >> --- a/libavcodec/hevc.h
> >> +++ b/libavcodec/hevc.h
> >> @@ -154,6 +154,9 @@ enum {
> >>      // get near that, though, so set a lower limit here with the maximum
> >>      // possible value for 4K video (at most 135 16x16 Ctb rows).
> >>      HEVC_MAX_ENTRY_POINT_OFFSETS = HEVC_MAX_TILE_COLUMNS * 135,
> >> +
> >> +    // A.3.7: Screen content coding extensions
> >> +    HEVC_MAX_PALETTE_PREDICTOR_SIZE = 128,
> >>  };
> >>
> >>
> >> diff --git a/libavcodec/hevc_ps.c b/libavcodec/hevc_ps.c
> >> index 584e2ba..7610d3b 100644
> >> --- a/libavcodec/hevc_ps.c
> >> +++ b/libavcodec/hevc_ps.c
> >> @@ -909,7 +909,7 @@ int ff_hevc_parse_sps(HEVCSPS *sps, GetBitContext *gb, unsigned int *sps_id,
> >>      HEVCWindow *ow;
> >>      int ret = 0;
> >>      int log2_diff_max_min_transform_block_size;
> >> -    int bit_depth_chroma, start, vui_present, sublayer_ordering_info;
> >> +    int bit_depth_chroma, start, vui_present, sublayer_ordering_info, num_comps;
> >>      int i;
> >>
> >>      // Coded parameters
> >> @@ -1130,8 +1130,11 @@ int ff_hevc_parse_sps(HEVCSPS *sps, GetBitContext *gb, unsigned int *sps_id,
> >>          decode_vui(gb, avctx, apply_defdispwin, sps);
> >>
> >>      if (get_bits1(gb)) { // sps_extension_flag
> >> -        sps->sps_range_extension_flag = get_bits1(gb);
> >> -        skip_bits(gb, 7); //sps_extension_7bits = get_bits(gb, 7);
> >> +        sps->sps_range_extension_flag      = get_bits1(gb);
> >> +        sps->sps_multilayer_extension_flag = get_bits1(gb);
> >> +        sps->sps_3d_extension_flag         = get_bits1(gb);
> >> +        sps->sps_scc_extension_flag        = get_bits1(gb);
> >> +        skip_bits(gb, 4); //sps_extension_4bits = get_bits(gb, 4);
> >>          if (sps->sps_range_extension_flag) {
> >>              sps->transform_skip_rotation_enabled_flag = get_bits1(gb);
> >>              sps->transform_skip_context_enabled_flag  = get_bits1(gb);
> >> @@ -1157,6 +1160,29 @@ int ff_hevc_parse_sps(HEVCSPS *sps, GetBitContext *gb, unsigned int *sps_id,
> >>                  av_log(avctx, AV_LOG_WARNING,
> >>                     "cabac_bypass_alignment_enabled_flag not yet implemented\n");
> >>          }
> >> +        if (sps->sps_multilayer_extension_flag || sps->sps_3d_extension_flag)
> >> +            av_log(avctx, AV_LOG_WARNING,
> >> +                   "multilayer_extension or 3d_extension not yet implemented\n");
>
> Also, you can't just ignore these. If you don't parse both, you'll be
> reading the wrong bits for SCC.
>
> Either add support for these two, or return AVERROR_PATCHWELCOME. Same
> for PPS.

Ah, should have noticed this.
Will fix, thx.

> >> +
> >> +        if (sps->sps_scc_extension_flag) {
> >> +            sps->sps_curr_pic_ref_enabled_flag = get_bits1(gb);
> >> +            sps->palette_mode_enabled_flag     = get_bits1(gb);
> >> +            if (sps->palette_mode_enabled_flag) {
> >> +                sps->palette_max_size = get_ue_golomb_long(gb);
> >> +                sps->delta_palette_max_predictor_size = get_ue_golomb_long(gb);
> >> +                sps->sps_palette_predictor_initializers_present_flag = get_bits1(gb);
> >> +
> >> +                if (sps->sps_palette_predictor_initializers_present_flag) {
> >> +                    sps->sps_num_palette_predictor_initializers_minus1 = get_ue_golomb_long(gb);
> >> +                    num_comps = sps->chroma_format_idc == 0 ? 1 : 3;
> >> +                    for (int comp = 0; comp < num_comps; comp++)
> >> +                        for (i = 0; i <= sps->sps_num_palette_predictor_initializers_minus1; i++)
> >> +                            sps->sps_palette_predictor_initializer[comp][i] = get_ue_golomb_long(gb);
> >
> > This isn't an ue value. Its length is a fixed value (each plane's
> > bitdepth, afaics).
> >
It's u(v), shall be in the range of 0 to (1 << BitDepth) - 1, inclusive, thx.

- Linjie


More information about the ffmpeg-devel mailing list