[FFmpeg-devel] [PATCH 2/4] avcodec/cbs_h265: add partial support for Multilayer extension fields in parameter set NALUs
James Almer
jamrial at gmail.com
Sat Jul 13 17:58:44 EEST 2024
Signed-off-by: James Almer <jamrial at gmail.com>
---
libavcodec/cbs_h2645.c | 3 +-
libavcodec/cbs_h265.h | 44 ++++++++++
libavcodec/cbs_h265_syntax_template.c | 113 +++++++++++++++++++++++++-
libavcodec/hevc/ps.h | 4 +-
4 files changed, 159 insertions(+), 5 deletions(-)
diff --git a/libavcodec/cbs_h2645.c b/libavcodec/cbs_h2645.c
index 5ec781ddab..828e56b8c0 100644
--- a/libavcodec/cbs_h2645.c
+++ b/libavcodec/cbs_h2645.c
@@ -499,7 +499,8 @@ static int cbs_h2645_fragment_add_nals(CodedBitstreamContext *ctx,
size_t size = nal->size;
enum AVCodecID codec_id = ctx->codec->codec_id;
- if (codec_id != AV_CODEC_ID_VVC && nal->nuh_layer_id > 0)
+ if (codec_id == AV_CODEC_ID_HEVC && nal->nuh_layer_id > 0 &&
+ (nal->type < HEVC_NAL_VPS || nal->type > HEVC_NAL_PPS))
continue;
// Remove trailing zeroes.
diff --git a/libavcodec/cbs_h265.h b/libavcodec/cbs_h265.h
index 91a5a55317..afb942ced5 100644
--- a/libavcodec/cbs_h265.h
+++ b/libavcodec/cbs_h265.h
@@ -248,12 +248,16 @@ typedef struct H265RawSPS {
uint8_t sps_video_parameter_set_id;
uint8_t sps_max_sub_layers_minus1;
+ uint8_t sps_ext_or_max_sub_layers_minus1;
uint8_t sps_temporal_id_nesting_flag;
H265RawProfileTierLevel profile_tier_level;
uint8_t sps_seq_parameter_set_id;
+ uint8_t update_rep_format_flag;
+ uint8_t sps_rep_format_idx;
+
uint8_t chroma_format_idc;
uint8_t separate_colour_plane_flag;
@@ -284,6 +288,8 @@ typedef struct H265RawSPS {
uint8_t max_transform_hierarchy_depth_intra;
uint8_t scaling_list_enabled_flag;
+ uint8_t sps_infer_scaling_list_flag;
+ uint8_t sps_scaling_list_ref_layer_id;
uint8_t sps_scaling_list_data_present_flag;
H265RawScalingList scaling_list;
@@ -342,6 +348,9 @@ typedef struct H265RawSPS {
uint8_t motion_vector_resolution_control_idc;
uint8_t intra_boundary_filtering_disable_flag;
+
+ // Multilayer extension.
+ uint8_t inter_view_mv_vert_constraint_flag;
} H265RawSPS;
typedef struct H265RawPPS {
@@ -433,6 +442,41 @@ typedef struct H265RawPPS {
uint8_t luma_bit_depth_entry_minus8;
uint8_t chroma_bit_depth_entry_minus8;
uint16_t pps_palette_predictor_initializers[3][128];
+
+ // Multilayer extension.
+ uint8_t poc_reset_info_present_flag;
+ uint8_t pps_infer_scaling_list_flag;
+ uint8_t pps_scaling_list_ref_layer_id;
+ uint8_t num_ref_loc_offsets;
+ uint8_t ref_loc_offset_layer_id[64];
+ uint8_t scaled_ref_layer_offset_present_flag[64];
+ int16_t scaled_ref_layer_left_offset[64];
+ int16_t scaled_ref_layer_top_offset[64];
+ int16_t scaled_ref_layer_right_offset[64];
+ int16_t scaled_ref_layer_bottom_offset[64];
+ uint8_t ref_region_offset_present_flag[64];
+ int16_t ref_region_left_offset[64];
+ int16_t ref_region_top_offset[64];
+ int16_t ref_region_right_offset[64];
+ int16_t ref_region_bottom_offset[64];
+ uint8_t resample_phase_set_present_flag[64];
+ uint8_t phase_hor_luma[64];
+ uint8_t phase_ver_luma[64];
+ uint8_t phase_hor_chroma_plus8[64];
+ uint8_t phase_ver_chroma_plus8[64];
+ uint8_t colour_mapping_enabled_flag;
+ uint8_t num_cm_ref_layers;
+ uint8_t cm_ref_layer_id[62];
+ uint8_t cm_octant_depth;
+ uint8_t cm_y_part_num_log2;
+ uint8_t luma_bit_depth_cm_input;
+ uint8_t chroma_bit_depth_cm_input;
+ uint8_t luma_bit_depth_cm_output;
+ uint8_t chroma_bit_depth_cm_output;
+ uint8_t cm_res_quant_bits;
+ uint8_t cm_delta_flc_bits;
+ int8_t cm_adapt_threshold_u_delta;
+ int8_t cm_adapt_threshold_v_delta;
} H265RawPPS;
typedef struct H265RawAUD {
diff --git a/libavcodec/cbs_h265_syntax_template.c b/libavcodec/cbs_h265_syntax_template.c
index c6db439b3b..2f8ee0db10 100644
--- a/libavcodec/cbs_h265_syntax_template.c
+++ b/libavcodec/cbs_h265_syntax_template.c
@@ -747,6 +747,16 @@ static int FUNC(sps_scc_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(vui_parameters_default)(CodedBitstreamContext *ctx,
RWContext *rw, H265RawVUI *current,
H265RawSPS *sps)
@@ -781,6 +791,7 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
int err, i;
unsigned int min_cb_log2_size_y, ctb_log2_size_y,
min_cb_size_y, min_tb_log2_size_y;
+ unsigned int multi_layer_ext_sps_flag;
HEADER("Sequence Parameter Set");
@@ -794,7 +805,17 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
return AVERROR_INVALIDDATA;
}
+ if (current->nal_unit_header.nuh_layer_id == 0)
u(3, sps_max_sub_layers_minus1, 0, vps->vps_max_sub_layers_minus1);
+ else {
+ u(3, sps_ext_or_max_sub_layers_minus1, 0, HEVC_MAX_SUB_LAYERS);
+ infer(sps_max_sub_layers_minus1, current->sps_ext_or_max_sub_layers_minus1 == 7
+ ? vps->vps_max_sub_layers_minus1
+ : current->sps_ext_or_max_sub_layers_minus1);
+ }
+ multi_layer_ext_sps_flag = current->nal_unit_header.nuh_layer_id &&
+ current->sps_ext_or_max_sub_layers_minus1 == 7;
+ if (!multi_layer_ext_sps_flag) {
flag(sps_temporal_id_nesting_flag);
if (vps->vps_temporal_id_nesting_flag &&
@@ -814,9 +835,20 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
CHECK(FUNC(profile_tier_level)(ctx, rw, ¤t->profile_tier_level,
1, current->sps_max_sub_layers_minus1));
+ } else {
+ if (current->sps_max_sub_layers_minus1 > 0)
+ infer(sps_temporal_id_nesting_flag, vps->vps_temporal_id_nesting_flag);
+ else
+ infer(sps_temporal_id_nesting_flag, 1);
+ }
ue(sps_seq_parameter_set_id, 0, 15);
+ if (multi_layer_ext_sps_flag) {
+ flag(update_rep_format_flag);
+ if (current->update_rep_format_flag)
+ ub(8, sps_rep_format_idx);
+ } else {
ue(chroma_format_idc, 0, 3);
if (current->chroma_format_idc == 3)
flag(separate_colour_plane_flag);
@@ -841,9 +873,11 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
ue(bit_depth_luma_minus8, 0, 8);
ue(bit_depth_chroma_minus8, 0, 8);
+ }
ue(log2_max_pic_order_cnt_lsb_minus4, 0, 12);
+ if (!multi_layer_ext_sps_flag) {
flag(sps_sub_layer_ordering_info_present_flag);
for (i = (current->sps_sub_layer_ordering_info_present_flag ?
0 : current->sps_max_sub_layers_minus1);
@@ -865,6 +899,7 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
current->sps_max_latency_increase_plus1[current->sps_max_sub_layers_minus1]);
}
}
+ }
ue(log2_min_luma_coding_block_size_minus3, 0, 3);
min_cb_log2_size_y = current->log2_min_luma_coding_block_size_minus3 + 3;
@@ -895,9 +930,17 @@ static int FUNC(sps)(CodedBitstreamContext *ctx, RWContext *rw,
flag(scaling_list_enabled_flag);
if (current->scaling_list_enabled_flag) {
+ if (multi_layer_ext_sps_flag)
+ flag(sps_infer_scaling_list_flag);
+ else
+ infer(sps_infer_scaling_list_flag, 0);
+ if (current->sps_infer_scaling_list_flag)
+ ub(6, sps_scaling_list_ref_layer_id);
+ else {
flag(sps_scaling_list_data_present_flag);
if (current->sps_scaling_list_data_present_flag)
CHECK(FUNC(scaling_list_data)(ctx, rw, ¤t->scaling_list));
+ }
} else {
infer(sps_scaling_list_data_present_flag, 0);
}
@@ -955,7 +998,7 @@ 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)
@@ -996,6 +1039,72 @@ static int FUNC(pps_range_extension)(CodedBitstreamContext *ctx, RWContext *rw,
return 0;
}
+static int FUNC(pps_multilayer_extension)(CodedBitstreamContext *ctx, RWContext *rw,
+ H265RawPPS *current)
+{
+ CodedBitstreamH265Context *h265 = ctx->priv_data;
+ const H265RawVPS *vps = h265->active_vps;
+ int offset;
+ int err, i;
+
+ flag(poc_reset_info_present_flag);
+ flag(pps_infer_scaling_list_flag);
+ if (current->pps_infer_scaling_list_flag)
+ ub(6, pps_scaling_list_ref_layer_id);
+
+ if (!vps) {
+ av_log(ctx->log_ctx, AV_LOG_ERROR, "VPS missing for PPS Multilayer Extension.\n");
+ return AVERROR_INVALIDDATA;
+ }
+
+ ue(num_ref_loc_offsets, 0, vps->vps_max_layers_minus1);
+ for (i = 0; i < current->num_ref_loc_offsets; i++) {
+ ubs(6, ref_loc_offset_layer_id[i], 1, i);
+ offset = current->ref_loc_offset_layer_id[i];
+ flags(scaled_ref_layer_offset_present_flag[i], 1, i);
+ if (current->scaled_ref_layer_offset_present_flag[i]) {
+ ses(scaled_ref_layer_left_offset[offset], -16384, 16383, 1, offset);
+ ses(scaled_ref_layer_top_offset[offset], -16384, 16383, 1, offset);
+ ses(scaled_ref_layer_right_offset[offset], -16384, 16383, 1, offset);
+ ses(scaled_ref_layer_bottom_offset[offset], -16384, 16383, 1, offset);
+ } else {
+ infer(scaled_ref_layer_left_offset[offset], 0);
+ infer(scaled_ref_layer_top_offset[offset], 0);
+ infer(scaled_ref_layer_right_offset[offset], 0);
+ infer(scaled_ref_layer_bottom_offset[offset], 0);
+ }
+ flags(ref_region_offset_present_flag[i], 1, i);
+ if (current->ref_region_offset_present_flag[i]) {
+ ses(ref_region_left_offset[offset], -16384, 16383, 1, offset);
+ ses(ref_region_top_offset[offset], -16384, 16383, 1, offset);
+ ses(ref_region_right_offset[offset], -16384, 16383, 1, offset);
+ ses(ref_region_bottom_offset[offset], -16384, 16383, 1, offset);
+ } else {
+ infer(ref_region_left_offset[offset], 0);
+ infer(ref_region_top_offset[offset], 0);
+ infer(ref_region_right_offset[offset], 0);
+ infer(ref_region_bottom_offset[offset], 0);
+ }
+ flags(resample_phase_set_present_flag[i], 1, i);
+ if (current->resample_phase_set_present_flag[i]) {
+ ues(phase_hor_luma[offset], 0, 31, 1, offset);
+ ues(phase_ver_luma[offset], 0, 31, 1, offset);
+ ues(phase_hor_chroma_plus8[offset], 0, 63, 1, offset);
+ ues(phase_ver_chroma_plus8[offset], 0, 63, 1, offset);
+ } else {
+ infer(phase_hor_luma[offset], 0);
+ infer(phase_ver_luma[offset], 0);
+ infer(phase_hor_chroma_plus8[offset], 8);
+ }
+ }
+
+ flag(colour_mapping_enabled_flag);
+ if (current->colour_mapping_enabled_flag)
+ return AVERROR_PATCHWELCOME;
+
+ return 0;
+}
+
static int FUNC(pps_scc_extension)(CodedBitstreamContext *ctx, RWContext *rw,
H265RawPPS *current)
{
@@ -1146,7 +1255,7 @@ static int FUNC(pps)(CodedBitstreamContext *ctx, RWContext *rw,
if (current->pps_range_extension_flag)
CHECK(FUNC(pps_range_extension)(ctx, rw, current));
if (current->pps_multilayer_extension_flag)
- return AVERROR_PATCHWELCOME;
+ CHECK(FUNC(pps_multilayer_extension)(ctx, rw, current));
if (current->pps_3d_extension_flag)
return AVERROR_PATCHWELCOME;
if (current->pps_scc_extension_flag)
diff --git a/libavcodec/hevc/ps.h b/libavcodec/hevc/ps.h
index 17395c5510..fab5a46273 100644
--- a/libavcodec/hevc/ps.h
+++ b/libavcodec/hevc/ps.h
@@ -390,8 +390,8 @@ typedef struct HEVCPPS {
uint8_t resample_phase_set_present_flag[64];
uint8_t phase_hor_luma[64];
uint8_t phase_ver_luma[64];
- int8_t phase_hor_chroma[64];
- int8_t phase_ver_chroma[64];
+ uint8_t phase_hor_chroma[64];
+ uint8_t phase_ver_chroma[64];
uint8_t colour_mapping_enabled_flag;
uint8_t num_cm_ref_layers;
uint8_t cm_ref_layer_id[62];
--
2.45.2
More information about the ffmpeg-devel
mailing list