xref: /third_party/ffmpeg/libavcodec/cbs_av1.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_AV1_H
20#define AVCODEC_CBS_AV1_H
21
22#include <stddef.h>
23#include <stdint.h>
24
25#include "av1.h"
26#include "cbs.h"
27
28
29typedef struct AV1RawOBUHeader {
30    uint8_t obu_forbidden_bit;
31    uint8_t obu_type;
32    uint8_t obu_extension_flag;
33    uint8_t obu_has_size_field;
34    uint8_t obu_reserved_1bit;
35
36    uint8_t temporal_id;
37    uint8_t spatial_id;
38    uint8_t extension_header_reserved_3bits;
39} AV1RawOBUHeader;
40
41typedef struct AV1RawColorConfig {
42    uint8_t high_bitdepth;
43    uint8_t twelve_bit;
44    uint8_t mono_chrome;
45
46    uint8_t color_description_present_flag;
47    uint8_t color_primaries;
48    uint8_t transfer_characteristics;
49    uint8_t matrix_coefficients;
50
51    uint8_t color_range;
52    uint8_t subsampling_x;
53    uint8_t subsampling_y;
54    uint8_t chroma_sample_position;
55    uint8_t separate_uv_delta_q;
56} AV1RawColorConfig;
57
58typedef struct AV1RawTimingInfo {
59    uint32_t num_units_in_display_tick;
60    uint32_t time_scale;
61
62    uint8_t equal_picture_interval;
63    uint32_t num_ticks_per_picture_minus_1;
64} AV1RawTimingInfo;
65
66typedef struct AV1RawDecoderModelInfo {
67    uint8_t  buffer_delay_length_minus_1;
68    uint32_t num_units_in_decoding_tick;
69    uint8_t  buffer_removal_time_length_minus_1;
70    uint8_t  frame_presentation_time_length_minus_1;
71} AV1RawDecoderModelInfo;
72
73typedef struct AV1RawSequenceHeader {
74    uint8_t seq_profile;
75    uint8_t still_picture;
76    uint8_t reduced_still_picture_header;
77
78    uint8_t timing_info_present_flag;
79    uint8_t decoder_model_info_present_flag;
80    uint8_t initial_display_delay_present_flag;
81    uint8_t operating_points_cnt_minus_1;
82
83    AV1RawTimingInfo       timing_info;
84    AV1RawDecoderModelInfo decoder_model_info;
85
86    uint16_t operating_point_idc[AV1_MAX_OPERATING_POINTS];
87    uint8_t  seq_level_idx[AV1_MAX_OPERATING_POINTS];
88    uint8_t  seq_tier[AV1_MAX_OPERATING_POINTS];
89    uint8_t  decoder_model_present_for_this_op[AV1_MAX_OPERATING_POINTS];
90    uint32_t decoder_buffer_delay[AV1_MAX_OPERATING_POINTS];
91    uint32_t encoder_buffer_delay[AV1_MAX_OPERATING_POINTS];
92    uint8_t  low_delay_mode_flag[AV1_MAX_OPERATING_POINTS];
93    uint8_t  initial_display_delay_present_for_this_op[AV1_MAX_OPERATING_POINTS];
94    uint8_t  initial_display_delay_minus_1[AV1_MAX_OPERATING_POINTS];
95
96    uint8_t  frame_width_bits_minus_1;
97    uint8_t  frame_height_bits_minus_1;
98    uint16_t max_frame_width_minus_1;
99    uint16_t max_frame_height_minus_1;
100
101    uint8_t frame_id_numbers_present_flag;
102    uint8_t delta_frame_id_length_minus_2;
103    uint8_t additional_frame_id_length_minus_1;
104
105    uint8_t use_128x128_superblock;
106    uint8_t enable_filter_intra;
107    uint8_t enable_intra_edge_filter;
108    uint8_t enable_interintra_compound;
109    uint8_t enable_masked_compound;
110    uint8_t enable_warped_motion;
111    uint8_t enable_dual_filter;
112
113    uint8_t enable_order_hint;
114    uint8_t enable_jnt_comp;
115    uint8_t enable_ref_frame_mvs;
116
117    uint8_t seq_choose_screen_content_tools;
118    uint8_t seq_force_screen_content_tools;
119    uint8_t seq_choose_integer_mv;
120    uint8_t seq_force_integer_mv;
121
122    uint8_t order_hint_bits_minus_1;
123
124    uint8_t enable_superres;
125    uint8_t enable_cdef;
126    uint8_t enable_restoration;
127
128    AV1RawColorConfig color_config;
129
130    uint8_t film_grain_params_present;
131} AV1RawSequenceHeader;
132
133typedef struct AV1RawFilmGrainParams {
134    uint8_t  apply_grain;
135    uint16_t grain_seed;
136    uint8_t  update_grain;
137    uint8_t  film_grain_params_ref_idx;
138    uint8_t  num_y_points;
139    uint8_t  point_y_value[14];
140    uint8_t  point_y_scaling[14];
141    uint8_t  chroma_scaling_from_luma;
142    uint8_t  num_cb_points;
143    uint8_t  point_cb_value[10];
144    uint8_t  point_cb_scaling[10];
145    uint8_t  num_cr_points;
146    uint8_t  point_cr_value[10];
147    uint8_t  point_cr_scaling[10];
148    uint8_t  grain_scaling_minus_8;
149    uint8_t  ar_coeff_lag;
150    uint8_t  ar_coeffs_y_plus_128[24];
151    uint8_t  ar_coeffs_cb_plus_128[25];
152    uint8_t  ar_coeffs_cr_plus_128[25];
153    uint8_t  ar_coeff_shift_minus_6;
154    uint8_t  grain_scale_shift;
155    uint8_t  cb_mult;
156    uint8_t  cb_luma_mult;
157    uint16_t cb_offset;
158    uint8_t  cr_mult;
159    uint8_t  cr_luma_mult;
160    uint16_t cr_offset;
161    uint8_t  overlap_flag;
162    uint8_t  clip_to_restricted_range;
163} AV1RawFilmGrainParams;
164
165typedef struct AV1RawFrameHeader {
166    uint8_t  show_existing_frame;
167    uint8_t  frame_to_show_map_idx;
168    uint32_t frame_presentation_time;
169    uint32_t display_frame_id;
170
171    uint8_t frame_type;
172    uint8_t show_frame;
173    uint8_t showable_frame;
174
175    uint8_t error_resilient_mode;
176    uint8_t disable_cdf_update;
177    uint8_t allow_screen_content_tools;
178    uint8_t force_integer_mv;
179
180    uint32_t current_frame_id;
181    uint8_t  frame_size_override_flag;
182    uint8_t  order_hint;
183
184    uint8_t  buffer_removal_time_present_flag;
185    uint32_t buffer_removal_time[AV1_MAX_OPERATING_POINTS];
186
187    uint8_t  primary_ref_frame;
188    uint16_t frame_width_minus_1;
189    uint16_t frame_height_minus_1;
190    uint8_t  use_superres;
191    uint8_t  coded_denom;
192    uint8_t  render_and_frame_size_different;
193    uint16_t render_width_minus_1;
194    uint16_t render_height_minus_1;
195
196    uint8_t found_ref[AV1_REFS_PER_FRAME];
197
198    uint8_t refresh_frame_flags;
199    uint8_t allow_intrabc;
200    uint8_t ref_order_hint[AV1_NUM_REF_FRAMES];
201    uint8_t frame_refs_short_signaling;
202    uint8_t last_frame_idx;
203    uint8_t golden_frame_idx;
204    int8_t  ref_frame_idx[AV1_REFS_PER_FRAME];
205    uint32_t delta_frame_id_minus1[AV1_REFS_PER_FRAME];
206
207    uint8_t allow_high_precision_mv;
208    uint8_t is_filter_switchable;
209    uint8_t interpolation_filter;
210    uint8_t is_motion_mode_switchable;
211    uint8_t use_ref_frame_mvs;
212
213    uint8_t disable_frame_end_update_cdf;
214
215    uint8_t uniform_tile_spacing_flag;
216    uint8_t tile_cols_log2;
217    uint8_t tile_rows_log2;
218    uint8_t width_in_sbs_minus_1[AV1_MAX_TILE_COLS];
219    uint8_t height_in_sbs_minus_1[AV1_MAX_TILE_ROWS];
220    uint16_t context_update_tile_id;
221    uint8_t tile_size_bytes_minus1;
222
223    // These are derived values, but it's very unhelpful to have to
224    // recalculate them all the time so we store them here.
225    uint16_t tile_cols;
226    uint16_t tile_rows;
227
228    uint8_t base_q_idx;
229    int8_t  delta_q_y_dc;
230    uint8_t diff_uv_delta;
231    int8_t  delta_q_u_dc;
232    int8_t  delta_q_u_ac;
233    int8_t  delta_q_v_dc;
234    int8_t  delta_q_v_ac;
235    uint8_t using_qmatrix;
236    uint8_t qm_y;
237    uint8_t qm_u;
238    uint8_t qm_v;
239
240    uint8_t segmentation_enabled;
241    uint8_t segmentation_update_map;
242    uint8_t segmentation_temporal_update;
243    uint8_t segmentation_update_data;
244    uint8_t feature_enabled[AV1_MAX_SEGMENTS][AV1_SEG_LVL_MAX];
245    int16_t feature_value[AV1_MAX_SEGMENTS][AV1_SEG_LVL_MAX];
246
247    uint8_t delta_q_present;
248    uint8_t delta_q_res;
249    uint8_t delta_lf_present;
250    uint8_t delta_lf_res;
251    uint8_t delta_lf_multi;
252
253    uint8_t loop_filter_level[4];
254    uint8_t loop_filter_sharpness;
255    uint8_t loop_filter_delta_enabled;
256    uint8_t loop_filter_delta_update;
257    uint8_t update_ref_delta[AV1_TOTAL_REFS_PER_FRAME];
258    int8_t  loop_filter_ref_deltas[AV1_TOTAL_REFS_PER_FRAME];
259    uint8_t update_mode_delta[2];
260    int8_t  loop_filter_mode_deltas[2];
261
262    uint8_t cdef_damping_minus_3;
263    uint8_t cdef_bits;
264    uint8_t cdef_y_pri_strength[8];
265    uint8_t cdef_y_sec_strength[8];
266    uint8_t cdef_uv_pri_strength[8];
267    uint8_t cdef_uv_sec_strength[8];
268
269    uint8_t lr_type[3];
270    uint8_t lr_unit_shift;
271    uint8_t lr_uv_shift;
272
273    uint8_t tx_mode;
274    uint8_t reference_select;
275    uint8_t skip_mode_present;
276
277    uint8_t allow_warped_motion;
278    uint8_t reduced_tx_set;
279
280    uint8_t is_global[AV1_TOTAL_REFS_PER_FRAME];
281    uint8_t is_rot_zoom[AV1_TOTAL_REFS_PER_FRAME];
282    uint8_t is_translation[AV1_TOTAL_REFS_PER_FRAME];
283    //AV1RawSubexp gm_params[AV1_TOTAL_REFS_PER_FRAME][6];
284    uint32_t gm_params[AV1_TOTAL_REFS_PER_FRAME][6];
285
286    AV1RawFilmGrainParams film_grain;
287} AV1RawFrameHeader;
288
289typedef struct AV1RawTileData {
290    uint8_t     *data;
291    AVBufferRef *data_ref;
292    size_t       data_size;
293} AV1RawTileData;
294
295typedef struct AV1RawTileGroup {
296    uint8_t  tile_start_and_end_present_flag;
297    uint16_t tg_start;
298    uint16_t tg_end;
299
300    AV1RawTileData tile_data;
301} AV1RawTileGroup;
302
303typedef struct AV1RawFrame {
304    AV1RawFrameHeader header;
305    AV1RawTileGroup   tile_group;
306} AV1RawFrame;
307
308typedef struct AV1RawTileList {
309    uint8_t output_frame_width_in_tiles_minus_1;
310    uint8_t output_frame_height_in_tiles_minus_1;
311    uint16_t tile_count_minus_1;
312
313    AV1RawTileData tile_data;
314} AV1RawTileList;
315
316typedef struct AV1RawMetadataHDRCLL {
317    uint16_t max_cll;
318    uint16_t max_fall;
319} AV1RawMetadataHDRCLL;
320
321typedef struct AV1RawMetadataHDRMDCV {
322    uint16_t primary_chromaticity_x[3];
323    uint16_t primary_chromaticity_y[3];
324    uint16_t white_point_chromaticity_x;
325    uint16_t white_point_chromaticity_y;
326    uint32_t luminance_max;
327    uint32_t luminance_min;
328} AV1RawMetadataHDRMDCV;
329
330typedef struct AV1RawMetadataScalability {
331    uint8_t scalability_mode_idc;
332    uint8_t spatial_layers_cnt_minus_1;
333    uint8_t spatial_layer_dimensions_present_flag;
334    uint8_t spatial_layer_description_present_flag;
335    uint8_t temporal_group_description_present_flag;
336    uint8_t scalability_structure_reserved_3bits;
337    uint16_t spatial_layer_max_width[4];
338    uint16_t spatial_layer_max_height[4];
339    uint8_t spatial_layer_ref_id[4];
340    uint8_t temporal_group_size;
341    uint8_t temporal_group_temporal_id[255];
342    uint8_t temporal_group_temporal_switching_up_point_flag[255];
343    uint8_t temporal_group_spatial_switching_up_point_flag[255];
344    uint8_t temporal_group_ref_cnt[255];
345    uint8_t temporal_group_ref_pic_diff[255][7];
346} AV1RawMetadataScalability;
347
348typedef struct AV1RawMetadataITUTT35 {
349    uint8_t itu_t_t35_country_code;
350    uint8_t itu_t_t35_country_code_extension_byte;
351
352    uint8_t     *payload;
353    AVBufferRef *payload_ref;
354    size_t       payload_size;
355} AV1RawMetadataITUTT35;
356
357typedef struct AV1RawMetadataTimecode {
358    uint8_t  counting_type;
359    uint8_t  full_timestamp_flag;
360    uint8_t  discontinuity_flag;
361    uint8_t  cnt_dropped_flag;
362    uint16_t n_frames;
363    uint8_t  seconds_value;
364    uint8_t  minutes_value;
365    uint8_t  hours_value;
366    uint8_t  seconds_flag;
367    uint8_t  minutes_flag;
368    uint8_t  hours_flag;
369    uint8_t  time_offset_length;
370    uint32_t time_offset_value;
371} AV1RawMetadataTimecode;
372
373typedef struct AV1RawMetadata {
374    uint64_t metadata_type;
375    union {
376        AV1RawMetadataHDRCLL      hdr_cll;
377        AV1RawMetadataHDRMDCV     hdr_mdcv;
378        AV1RawMetadataScalability scalability;
379        AV1RawMetadataITUTT35     itut_t35;
380        AV1RawMetadataTimecode    timecode;
381    } metadata;
382} AV1RawMetadata;
383
384typedef struct AV1RawPadding {
385    uint8_t     *payload;
386    AVBufferRef *payload_ref;
387    size_t       payload_size;
388} AV1RawPadding;
389
390
391typedef struct AV1RawOBU {
392    AV1RawOBUHeader header;
393
394    size_t obu_size;
395
396    union {
397        AV1RawSequenceHeader sequence_header;
398        AV1RawFrameHeader    frame_header;
399        AV1RawFrame          frame;
400        AV1RawTileGroup      tile_group;
401        AV1RawTileList       tile_list;
402        AV1RawMetadata       metadata;
403        AV1RawPadding        padding;
404    } obu;
405} AV1RawOBU;
406
407typedef struct AV1ReferenceFrameState {
408    int valid;          // RefValid
409    int frame_id;       // RefFrameId
410    int upscaled_width; // RefUpscaledWidth
411    int frame_width;    // RefFrameWidth
412    int frame_height;   // RefFrameHeight
413    int render_width;   // RefRenderWidth
414    int render_height;  // RefRenderHeight
415    int frame_type;     // RefFrameType
416    int subsampling_x;  // RefSubsamplingX
417    int subsampling_y;  // RefSubsamplingY
418    int bit_depth;      // RefBitDepth
419    int order_hint;     // RefOrderHint
420
421    int8_t  loop_filter_ref_deltas[AV1_TOTAL_REFS_PER_FRAME];
422    int8_t  loop_filter_mode_deltas[2];
423    uint8_t feature_enabled[AV1_MAX_SEGMENTS][AV1_SEG_LVL_MAX];
424    int16_t feature_value[AV1_MAX_SEGMENTS][AV1_SEG_LVL_MAX];
425} AV1ReferenceFrameState;
426
427typedef struct CodedBitstreamAV1Context {
428    const AVClass *class;
429
430    AV1RawSequenceHeader *sequence_header;
431    AVBufferRef          *sequence_header_ref;
432
433    int     seen_frame_header;
434    AVBufferRef *frame_header_ref;
435    uint8_t     *frame_header;
436    size_t       frame_header_size;
437
438    int temporal_id;
439    int spatial_id;
440    int operating_point_idc;
441
442    int bit_depth;
443    int order_hint;
444    int frame_width;
445    int frame_height;
446    int upscaled_width;
447    int render_width;
448    int render_height;
449
450    int num_planes;
451    int coded_lossless;
452    int all_lossless;
453    int tile_cols;
454    int tile_rows;
455    int tile_num;
456
457    AV1ReferenceFrameState ref[AV1_NUM_REF_FRAMES];
458
459    // AVOptions
460    int operating_point;
461} CodedBitstreamAV1Context;
462
463
464#endif /* AVCODEC_CBS_AV1_H */
465