xref: /third_party/ffmpeg/libavcodec/cbs_h264.h (revision cabdff1a)
1/*
2 * This file is part of FFmpeg.
3 *
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19#ifndef AVCODEC_CBS_H264_H
20#define AVCODEC_CBS_H264_H
21
22#include <stddef.h>
23#include <stdint.h>
24
25#include "cbs.h"
26#include "cbs_h2645.h"
27#include "cbs_sei.h"
28#include "h264.h"
29
30
31typedef struct H264RawNALUnitHeader {
32    uint8_t nal_ref_idc;
33    uint8_t nal_unit_type;
34
35    uint8_t svc_extension_flag;
36    uint8_t avc_3d_extension_flag;
37} H264RawNALUnitHeader;
38
39typedef struct H264RawScalingList {
40    int8_t delta_scale[64];
41} H264RawScalingList;
42
43typedef struct H264RawHRD {
44    uint8_t cpb_cnt_minus1;
45    uint8_t bit_rate_scale;
46    uint8_t cpb_size_scale;
47
48    uint32_t bit_rate_value_minus1[H264_MAX_CPB_CNT];
49    uint32_t cpb_size_value_minus1[H264_MAX_CPB_CNT];
50    uint8_t cbr_flag[H264_MAX_CPB_CNT];
51
52    uint8_t initial_cpb_removal_delay_length_minus1;
53    uint8_t cpb_removal_delay_length_minus1;
54    uint8_t dpb_output_delay_length_minus1;
55    uint8_t time_offset_length;
56} H264RawHRD;
57
58typedef struct H264RawVUI {
59    uint8_t aspect_ratio_info_present_flag;
60    uint8_t aspect_ratio_idc;
61    uint16_t sar_width;
62    uint16_t sar_height;
63
64    uint8_t overscan_info_present_flag;
65    uint8_t overscan_appropriate_flag;
66
67    uint8_t video_signal_type_present_flag;
68    uint8_t video_format;
69    uint8_t video_full_range_flag;
70    uint8_t colour_description_present_flag;
71    uint8_t colour_primaries;
72    uint8_t transfer_characteristics;
73    uint8_t matrix_coefficients;
74
75    uint8_t chroma_loc_info_present_flag;
76    uint8_t chroma_sample_loc_type_top_field;
77    uint8_t chroma_sample_loc_type_bottom_field;
78
79    uint8_t timing_info_present_flag;
80    uint32_t num_units_in_tick;
81    uint32_t time_scale;
82    uint8_t fixed_frame_rate_flag;
83
84    uint8_t nal_hrd_parameters_present_flag;
85    H264RawHRD nal_hrd_parameters;
86    uint8_t vcl_hrd_parameters_present_flag;
87    H264RawHRD vcl_hrd_parameters;
88    uint8_t low_delay_hrd_flag;
89
90    uint8_t pic_struct_present_flag;
91
92    uint8_t bitstream_restriction_flag;
93    uint8_t motion_vectors_over_pic_boundaries_flag;
94    uint8_t max_bytes_per_pic_denom;
95    uint8_t max_bits_per_mb_denom;
96    uint8_t log2_max_mv_length_horizontal;
97    uint8_t log2_max_mv_length_vertical;
98    uint8_t max_num_reorder_frames;
99    uint8_t max_dec_frame_buffering;
100} H264RawVUI;
101
102typedef struct H264RawSPS {
103    H264RawNALUnitHeader nal_unit_header;
104
105    uint8_t profile_idc;
106    uint8_t constraint_set0_flag;
107    uint8_t constraint_set1_flag;
108    uint8_t constraint_set2_flag;
109    uint8_t constraint_set3_flag;
110    uint8_t constraint_set4_flag;
111    uint8_t constraint_set5_flag;
112    uint8_t reserved_zero_2bits;
113    uint8_t level_idc;
114
115    uint8_t seq_parameter_set_id;
116
117    uint8_t chroma_format_idc;
118    uint8_t separate_colour_plane_flag;
119    uint8_t bit_depth_luma_minus8;
120    uint8_t bit_depth_chroma_minus8;
121    uint8_t qpprime_y_zero_transform_bypass_flag;
122
123    uint8_t seq_scaling_matrix_present_flag;
124    uint8_t seq_scaling_list_present_flag[12];
125    H264RawScalingList scaling_list_4x4[6];
126    H264RawScalingList scaling_list_8x8[6];
127
128    uint8_t log2_max_frame_num_minus4;
129    uint8_t pic_order_cnt_type;
130    uint8_t log2_max_pic_order_cnt_lsb_minus4;
131    uint8_t delta_pic_order_always_zero_flag;
132    int32_t offset_for_non_ref_pic;
133    int32_t offset_for_top_to_bottom_field;
134    uint8_t num_ref_frames_in_pic_order_cnt_cycle;
135    int32_t offset_for_ref_frame[256];
136
137    uint8_t max_num_ref_frames;
138    uint8_t gaps_in_frame_num_allowed_flag;
139
140    uint16_t pic_width_in_mbs_minus1;
141    uint16_t pic_height_in_map_units_minus1;
142
143    uint8_t frame_mbs_only_flag;
144    uint8_t mb_adaptive_frame_field_flag;
145    uint8_t direct_8x8_inference_flag;
146
147    uint8_t frame_cropping_flag;
148    uint16_t frame_crop_left_offset;
149    uint16_t frame_crop_right_offset;
150    uint16_t frame_crop_top_offset;
151    uint16_t frame_crop_bottom_offset;
152
153    uint8_t vui_parameters_present_flag;
154    H264RawVUI vui;
155} H264RawSPS;
156
157typedef struct H264RawSPSExtension {
158    H264RawNALUnitHeader nal_unit_header;
159
160    uint8_t seq_parameter_set_id;
161
162    uint8_t aux_format_idc;
163    uint8_t bit_depth_aux_minus8;
164    uint8_t alpha_incr_flag;
165    uint16_t alpha_opaque_value;
166    uint16_t alpha_transparent_value;
167
168    uint8_t additional_extension_flag;
169} H264RawSPSExtension;
170
171typedef struct H264RawPPS {
172    H264RawNALUnitHeader nal_unit_header;
173
174    uint8_t pic_parameter_set_id;
175    uint8_t seq_parameter_set_id;
176
177    uint8_t entropy_coding_mode_flag;
178    uint8_t bottom_field_pic_order_in_frame_present_flag;
179
180    uint8_t num_slice_groups_minus1;
181    uint8_t slice_group_map_type;
182    uint16_t run_length_minus1[H264_MAX_SLICE_GROUPS];
183    uint16_t top_left[H264_MAX_SLICE_GROUPS];
184    uint16_t bottom_right[H264_MAX_SLICE_GROUPS];
185    uint8_t slice_group_change_direction_flag;
186    uint16_t slice_group_change_rate_minus1;
187    uint16_t pic_size_in_map_units_minus1;
188
189    uint8_t *slice_group_id;
190    AVBufferRef *slice_group_id_ref;
191
192    uint8_t num_ref_idx_l0_default_active_minus1;
193    uint8_t num_ref_idx_l1_default_active_minus1;
194
195    uint8_t weighted_pred_flag;
196    uint8_t weighted_bipred_idc;
197
198    int8_t pic_init_qp_minus26;
199    int8_t pic_init_qs_minus26;
200    int8_t chroma_qp_index_offset;
201
202    uint8_t deblocking_filter_control_present_flag;
203    uint8_t constrained_intra_pred_flag;
204
205    uint8_t more_rbsp_data;
206
207    uint8_t redundant_pic_cnt_present_flag;
208    uint8_t transform_8x8_mode_flag;
209
210    uint8_t pic_scaling_matrix_present_flag;
211    uint8_t pic_scaling_list_present_flag[12];
212    H264RawScalingList scaling_list_4x4[6];
213    H264RawScalingList scaling_list_8x8[6];
214
215    int8_t second_chroma_qp_index_offset;
216} H264RawPPS;
217
218typedef struct H264RawAUD {
219    H264RawNALUnitHeader nal_unit_header;
220
221    uint8_t primary_pic_type;
222} H264RawAUD;
223
224typedef struct H264RawSEIBufferingPeriod {
225    uint8_t seq_parameter_set_id;
226    struct {
227        uint32_t initial_cpb_removal_delay[H264_MAX_CPB_CNT];
228        uint32_t initial_cpb_removal_delay_offset[H264_MAX_CPB_CNT];
229    } nal, vcl;
230} H264RawSEIBufferingPeriod;
231
232typedef struct H264RawSEIPicTimestamp {
233    uint8_t ct_type;
234    uint8_t nuit_field_based_flag;
235    uint8_t counting_type;
236    uint8_t full_timestamp_flag;
237    uint8_t discontinuity_flag;
238    uint8_t cnt_dropped_flag;
239    uint8_t n_frames;
240    uint8_t seconds_flag;
241    uint8_t seconds_value;
242    uint8_t minutes_flag;
243    uint8_t minutes_value;
244    uint8_t hours_flag;
245    uint8_t hours_value;
246    int32_t time_offset;
247} H264RawSEIPicTimestamp;
248
249typedef struct H264RawSEIPicTiming {
250    uint32_t cpb_removal_delay;
251    uint32_t dpb_output_delay;
252    uint8_t pic_struct;
253    uint8_t clock_timestamp_flag[3];
254    H264RawSEIPicTimestamp timestamp[3];
255} H264RawSEIPicTiming;
256
257typedef struct H264RawSEIPanScanRect {
258    uint32_t pan_scan_rect_id;
259    uint8_t  pan_scan_rect_cancel_flag;
260    uint8_t  pan_scan_cnt_minus1;
261    int32_t  pan_scan_rect_left_offset[3];
262    int32_t  pan_scan_rect_right_offset[3];
263    int32_t  pan_scan_rect_top_offset[3];
264    int32_t  pan_scan_rect_bottom_offset[3];
265    uint16_t pan_scan_rect_repetition_period;
266} H264RawSEIPanScanRect;
267
268typedef struct H264RawSEIRecoveryPoint {
269    uint16_t recovery_frame_cnt;
270    uint8_t exact_match_flag;
271    uint8_t broken_link_flag;
272    uint8_t changing_slice_group_idc;
273} H264RawSEIRecoveryPoint;
274
275typedef struct H264RawFilmGrainCharacteristics {
276    uint8_t      film_grain_characteristics_cancel_flag;
277    uint8_t      film_grain_model_id;
278    uint8_t      separate_colour_description_present_flag;
279    uint8_t      film_grain_bit_depth_luma_minus8;
280    uint8_t      film_grain_bit_depth_chroma_minus8;
281    uint8_t      film_grain_full_range_flag;
282    uint8_t      film_grain_colour_primaries;
283    uint8_t      film_grain_transfer_characteristics;
284    uint8_t      film_grain_matrix_coefficients;
285    uint8_t      blending_mode_id;
286    uint8_t      log2_scale_factor;
287    uint8_t      comp_model_present_flag[3];
288    uint8_t      num_intensity_intervals_minus1[3];
289    uint8_t      num_model_values_minus1[3];
290    uint8_t      intensity_interval_lower_bound[3][256];
291    uint8_t      intensity_interval_upper_bound[3][256];
292    int16_t      comp_model_value[3][256][6];
293    uint8_t      film_grain_characteristics_repetition_period;
294} H264RawFilmGrainCharacteristics;
295
296typedef struct H264RawSEIDisplayOrientation {
297    uint8_t display_orientation_cancel_flag;
298    uint8_t hor_flip;
299    uint8_t ver_flip;
300    uint16_t anticlockwise_rotation;
301    uint16_t display_orientation_repetition_period;
302    uint8_t display_orientation_extension_flag;
303} H264RawSEIDisplayOrientation;
304
305typedef struct H264RawSEI {
306    H264RawNALUnitHeader nal_unit_header;
307    SEIRawMessageList    message_list;
308} H264RawSEI;
309
310typedef struct H264RawSliceHeader {
311    H264RawNALUnitHeader nal_unit_header;
312
313    uint32_t first_mb_in_slice;
314    uint8_t slice_type;
315
316    uint8_t pic_parameter_set_id;
317
318    uint8_t colour_plane_id;
319
320    uint16_t frame_num;
321    uint8_t field_pic_flag;
322    uint8_t bottom_field_flag;
323
324    uint16_t idr_pic_id;
325
326    uint16_t pic_order_cnt_lsb;
327    int32_t delta_pic_order_cnt_bottom;
328    int32_t delta_pic_order_cnt[2];
329
330    uint8_t redundant_pic_cnt;
331    uint8_t direct_spatial_mv_pred_flag;
332
333    uint8_t num_ref_idx_active_override_flag;
334    uint8_t num_ref_idx_l0_active_minus1;
335    uint8_t num_ref_idx_l1_active_minus1;
336
337    uint8_t ref_pic_list_modification_flag_l0;
338    uint8_t ref_pic_list_modification_flag_l1;
339    struct {
340        uint8_t modification_of_pic_nums_idc;
341        int32_t abs_diff_pic_num_minus1;
342        uint8_t long_term_pic_num;
343    } rplm_l0[H264_MAX_RPLM_COUNT], rplm_l1[H264_MAX_RPLM_COUNT];
344
345    uint8_t luma_log2_weight_denom;
346    uint8_t chroma_log2_weight_denom;
347
348    uint8_t luma_weight_l0_flag[H264_MAX_REFS];
349    int8_t luma_weight_l0[H264_MAX_REFS];
350    int8_t luma_offset_l0[H264_MAX_REFS];
351    uint8_t chroma_weight_l0_flag[H264_MAX_REFS];
352    int8_t chroma_weight_l0[H264_MAX_REFS][2];
353    int8_t chroma_offset_l0[H264_MAX_REFS][2];
354
355    uint8_t luma_weight_l1_flag[H264_MAX_REFS];
356    int8_t luma_weight_l1[H264_MAX_REFS];
357    int8_t luma_offset_l1[H264_MAX_REFS];
358    uint8_t chroma_weight_l1_flag[H264_MAX_REFS];
359    int8_t chroma_weight_l1[H264_MAX_REFS][2];
360    int8_t chroma_offset_l1[H264_MAX_REFS][2];
361
362    uint8_t no_output_of_prior_pics_flag;
363    uint8_t long_term_reference_flag;
364
365    uint8_t adaptive_ref_pic_marking_mode_flag;
366    struct {
367        uint8_t memory_management_control_operation;
368        int32_t difference_of_pic_nums_minus1;
369        uint8_t long_term_pic_num;
370        uint8_t long_term_frame_idx;
371        uint8_t max_long_term_frame_idx_plus1;
372    } mmco[H264_MAX_MMCO_COUNT];
373
374    uint8_t cabac_init_idc;
375
376    int8_t slice_qp_delta;
377
378    uint8_t sp_for_switch_flag;
379    int8_t slice_qs_delta;
380
381    uint8_t disable_deblocking_filter_idc;
382    int8_t slice_alpha_c0_offset_div2;
383    int8_t slice_beta_offset_div2;
384
385    uint16_t slice_group_change_cycle;
386} H264RawSliceHeader;
387
388typedef struct H264RawSlice {
389    H264RawSliceHeader header;
390
391    uint8_t     *data;
392    AVBufferRef *data_ref;
393    size_t       data_size;
394    int          data_bit_start;
395} H264RawSlice;
396
397typedef struct H264RawFiller {
398    H264RawNALUnitHeader nal_unit_header;
399
400    uint32_t filler_size;
401} H264RawFiller;
402
403
404typedef struct CodedBitstreamH264Context {
405    // Reader/writer context in common with the H.265 implementation.
406    CodedBitstreamH2645Context common;
407
408    // All currently available parameter sets.  These are updated when
409    // any parameter set NAL unit is read/written with this context.
410    AVBufferRef *sps_ref[H264_MAX_SPS_COUNT];
411    AVBufferRef *pps_ref[H264_MAX_PPS_COUNT];
412    H264RawSPS *sps[H264_MAX_SPS_COUNT];
413    H264RawPPS *pps[H264_MAX_PPS_COUNT];
414
415    // The currently active parameter sets.  These are updated when any
416    // NAL unit refers to the relevant parameter set.  These pointers
417    // must also be present in the arrays above.
418    const H264RawSPS *active_sps;
419    const H264RawPPS *active_pps;
420
421    // The NAL unit type of the most recent normal slice.  This is required
422    // to be able to read/write auxiliary slices, because IdrPicFlag is
423    // otherwise unknown.
424    uint8_t last_slice_nal_unit_type;
425} CodedBitstreamH264Context;
426
427#endif /* AVCODEC_CBS_H264_H */
428