Commit | Line | Data |
---|---|---|
26b4fe82 AJ |
1 | /* |
2 | * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder | |
3 | * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at> | |
4 | * | |
5 | * This file is part of FFmpeg. | |
6 | * | |
7 | * FFmpeg is free software; you can redistribute it and/or | |
8 | * modify it under the terms of the GNU Lesser General Public | |
9 | * License as published by the Free Software Foundation; either | |
10 | * version 2.1 of the License, or (at your option) any later version. | |
11 | * | |
12 | * FFmpeg is distributed in the hope that it will be useful, | |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
15 | * Lesser General Public License for more details. | |
16 | * | |
17 | * You should have received a copy of the GNU Lesser General Public | |
18 | * License along with FFmpeg; if not, write to the Free Software | |
19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
26b4fe82 AJ |
20 | */ |
21 | ||
22 | /** | |
23 | * @file h264.h | |
24 | * H.264 / AVC / MPEG4 part10 codec. | |
25 | * @author Michael Niedermayer <michaelni@gmx.at> | |
26 | */ | |
27 | ||
98790382 SS |
28 | #ifndef AVCODEC_H264_H |
29 | #define AVCODEC_H264_H | |
26b4fe82 AJ |
30 | |
31 | #include "dsputil.h" | |
32 | #include "cabac.h" | |
33 | #include "mpegvideo.h" | |
c92a30bb | 34 | #include "h264pred.h" |
26b4fe82 AJ |
35 | |
36 | #define interlaced_dct interlaced_dct_is_a_bad_name | |
90b5b51e | 37 | #define mb_intra mb_intra_is_not_initialized_see_mb_type |
26b4fe82 AJ |
38 | |
39 | #define LUMA_DC_BLOCK_INDEX 25 | |
40 | #define CHROMA_DC_BLOCK_INDEX 26 | |
41 | ||
42 | #define CHROMA_DC_COEFF_TOKEN_VLC_BITS 8 | |
43 | #define COEFF_TOKEN_VLC_BITS 8 | |
44 | #define TOTAL_ZEROS_VLC_BITS 9 | |
45 | #define CHROMA_DC_TOTAL_ZEROS_VLC_BITS 3 | |
46 | #define RUN_VLC_BITS 3 | |
47 | #define RUN7_VLC_BITS 6 | |
48 | ||
49 | #define MAX_SPS_COUNT 32 | |
50 | #define MAX_PPS_COUNT 256 | |
51 | ||
52 | #define MAX_MMCO_COUNT 66 | |
53 | ||
64b9d48f AS |
54 | #define MAX_DELAYED_PIC_COUNT 16 |
55 | ||
26b4fe82 AJ |
56 | /* Compiling in interlaced support reduces the speed |
57 | * of progressive decoding by about 2%. */ | |
58 | #define ALLOW_INTERLACE | |
59 | ||
0af6967e MN |
60 | #define ALLOW_NOCHROMA |
61 | ||
b735aeea MN |
62 | /** |
63 | * The maximum number of slices supported by the decoder. | |
64 | * must be a power of 2 | |
65 | */ | |
66 | #define MAX_SLICES 16 | |
67 | ||
26b4fe82 AJ |
68 | #ifdef ALLOW_INTERLACE |
69 | #define MB_MBAFF h->mb_mbaff | |
70 | #define MB_FIELD h->mb_field_decoding_flag | |
71 | #define FRAME_MBAFF h->mb_aff_frame | |
d6c52130 | 72 | #define FIELD_PICTURE (s->picture_structure != PICT_FRAME) |
26b4fe82 AJ |
73 | #else |
74 | #define MB_MBAFF 0 | |
75 | #define MB_FIELD 0 | |
76 | #define FRAME_MBAFF 0 | |
bbb3edb8 | 77 | #define FIELD_PICTURE 0 |
26b4fe82 AJ |
78 | #undef IS_INTERLACED |
79 | #define IS_INTERLACED(mb_type) 0 | |
80 | #endif | |
f3e53d9f | 81 | #define FIELD_OR_MBAFF_PICTURE (FRAME_MBAFF || FIELD_PICTURE) |
26b4fe82 | 82 | |
0af6967e MN |
83 | #ifdef ALLOW_NOCHROMA |
84 | #define CHROMA h->sps.chroma_format_idc | |
85 | #else | |
86 | #define CHROMA 1 | |
87 | #endif | |
88 | ||
fedec603 AS |
89 | #ifndef ENABLE_H264_ENCODER |
90 | #define ENABLE_H264_ENCODER 0 | |
91 | #endif | |
92 | ||
44ee6135 MN |
93 | #define EXTENDED_SAR 255 |
94 | ||
95 | /* NAL unit types */ | |
96 | enum { | |
2791c608 MN |
97 | NAL_SLICE=1, |
98 | NAL_DPA, | |
99 | NAL_DPB, | |
100 | NAL_DPC, | |
101 | NAL_IDR_SLICE, | |
102 | NAL_SEI, | |
103 | NAL_SPS, | |
104 | NAL_PPS, | |
105 | NAL_AUD, | |
106 | NAL_END_SEQUENCE, | |
107 | NAL_END_STREAM, | |
108 | NAL_FILLER_DATA, | |
109 | NAL_SPS_EXT, | |
110 | NAL_AUXILIARY_SLICE=19 | |
44ee6135 MN |
111 | }; |
112 | ||
26b4fe82 AJ |
113 | /** |
114 | * Sequence parameter set | |
115 | */ | |
116 | typedef struct SPS{ | |
117 | ||
118 | int profile_idc; | |
119 | int level_idc; | |
0af6967e | 120 | int chroma_format_idc; |
26b4fe82 AJ |
121 | int transform_bypass; ///< qpprime_y_zero_transform_bypass_flag |
122 | int log2_max_frame_num; ///< log2_max_frame_num_minus4 + 4 | |
123 | int poc_type; ///< pic_order_cnt_type | |
124 | int log2_max_poc_lsb; ///< log2_max_pic_order_cnt_lsb_minus4 | |
125 | int delta_pic_order_always_zero_flag; | |
126 | int offset_for_non_ref_pic; | |
127 | int offset_for_top_to_bottom_field; | |
128 | int poc_cycle_length; ///< num_ref_frames_in_pic_order_cnt_cycle | |
129 | int ref_frame_count; ///< num_ref_frames | |
130 | int gaps_in_frame_num_allowed_flag; | |
bcc3476c JD |
131 | int mb_width; ///< pic_width_in_mbs_minus1 + 1 |
132 | int mb_height; ///< pic_height_in_map_units_minus1 + 1 | |
26b4fe82 AJ |
133 | int frame_mbs_only_flag; |
134 | int mb_aff; ///<mb_adaptive_frame_field_flag | |
135 | int direct_8x8_inference_flag; | |
136 | int crop; ///< frame_cropping_flag | |
59eeddf0 MN |
137 | unsigned int crop_left; ///< frame_cropping_rect_left_offset |
138 | unsigned int crop_right; ///< frame_cropping_rect_right_offset | |
139 | unsigned int crop_top; ///< frame_cropping_rect_top_offset | |
140 | unsigned int crop_bottom; ///< frame_cropping_rect_bottom_offset | |
26b4fe82 AJ |
141 | int vui_parameters_present_flag; |
142 | AVRational sar; | |
143 | int timing_info_present_flag; | |
144 | uint32_t num_units_in_tick; | |
145 | uint32_t time_scale; | |
146 | int fixed_frame_rate_flag; | |
147 | short offset_for_ref_frame[256]; //FIXME dyn aloc? | |
148 | int bitstream_restriction_flag; | |
149 | int num_reorder_frames; | |
150 | int scaling_matrix_present; | |
151 | uint8_t scaling_matrix4[6][16]; | |
152 | uint8_t scaling_matrix8[2][64]; | |
153 | }SPS; | |
154 | ||
155 | /** | |
156 | * Picture parameter set | |
157 | */ | |
158 | typedef struct PPS{ | |
159 | unsigned int sps_id; | |
160 | int cabac; ///< entropy_coding_mode_flag | |
161 | int pic_order_present; ///< pic_order_present_flag | |
162 | int slice_group_count; ///< num_slice_groups_minus1 + 1 | |
163 | int mb_slice_group_map_type; | |
164 | unsigned int ref_count[2]; ///< num_ref_idx_l0/1_active_minus1 + 1 | |
165 | int weighted_pred; ///< weighted_pred_flag | |
166 | int weighted_bipred_idc; | |
167 | int init_qp; ///< pic_init_qp_minus26 + 26 | |
168 | int init_qs; ///< pic_init_qs_minus26 + 26 | |
4691a77d | 169 | int chroma_qp_index_offset[2]; |
26b4fe82 AJ |
170 | int deblocking_filter_parameters_present; ///< deblocking_filter_parameters_present_flag |
171 | int constrained_intra_pred; ///< constrained_intra_pred_flag | |
172 | int redundant_pic_cnt_present; ///< redundant_pic_cnt_present_flag | |
173 | int transform_8x8_mode; ///< transform_8x8_mode_flag | |
174 | uint8_t scaling_matrix4[6][16]; | |
175 | uint8_t scaling_matrix8[2][64]; | |
5a78bfbd | 176 | uint8_t chroma_qp_table[2][64]; ///< pre-scaled (with chroma_qp_index_offset) version of qp_table |
4691a77d | 177 | int chroma_qp_diff; |
26b4fe82 AJ |
178 | }PPS; |
179 | ||
180 | /** | |
181 | * Memory management control operation opcode. | |
182 | */ | |
183 | typedef enum MMCOOpcode{ | |
184 | MMCO_END=0, | |
185 | MMCO_SHORT2UNUSED, | |
186 | MMCO_LONG2UNUSED, | |
187 | MMCO_SHORT2LONG, | |
188 | MMCO_SET_MAX_LONG, | |
189 | MMCO_RESET, | |
190 | MMCO_LONG, | |
191 | } MMCOOpcode; | |
192 | ||
193 | /** | |
194 | * Memory management control operation. | |
195 | */ | |
196 | typedef struct MMCO{ | |
197 | MMCOOpcode opcode; | |
0d0447ea JD |
198 | int short_pic_num; ///< pic_num without wrapping (pic_num & max_pic_num) |
199 | int long_arg; ///< index, pic_num, or num long refs depending on opcode | |
26b4fe82 AJ |
200 | } MMCO; |
201 | ||
202 | /** | |
203 | * H264Context | |
204 | */ | |
205 | typedef struct H264Context{ | |
206 | MpegEncContext s; | |
207 | int nal_ref_idc; | |
208 | int nal_unit_type; | |
24456882 AÖ |
209 | uint8_t *rbsp_buffer[2]; |
210 | unsigned int rbsp_buffer_size[2]; | |
26b4fe82 AJ |
211 | |
212 | /** | |
213 | * Used to parse AVC variant of h264 | |
214 | */ | |
215 | int is_avc; ///< this flag is != 0 if codec is avc1 | |
216 | int got_avcC; ///< flag used to parse avcC data only once | |
217 | int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4) | |
218 | ||
4691a77d | 219 | int chroma_qp[2]; //QPc |
26b4fe82 AJ |
220 | |
221 | int prev_mb_skipped; | |
222 | int next_mb_skipped; | |
223 | ||
224 | //prediction stuff | |
225 | int chroma_pred_mode; | |
226 | int intra16x16_pred_mode; | |
227 | ||
228 | int top_mb_xy; | |
229 | int left_mb_xy[2]; | |
230 | ||
231 | int8_t intra4x4_pred_mode_cache[5*8]; | |
232 | int8_t (*intra4x4_pred_mode)[8]; | |
c92a30bb | 233 | H264PredContext hpc; |
26b4fe82 AJ |
234 | unsigned int topleft_samples_available; |
235 | unsigned int top_samples_available; | |
236 | unsigned int topright_samples_available; | |
237 | unsigned int left_samples_available; | |
238 | uint8_t (*top_borders[2])[16+2*8]; | |
239 | uint8_t left_border[2*(17+2*9)]; | |
240 | ||
241 | /** | |
242 | * non zero coeff count cache. | |
243 | * is 64 if not available. | |
244 | */ | |
245 | DECLARE_ALIGNED_8(uint8_t, non_zero_count_cache[6*8]); | |
246 | uint8_t (*non_zero_count)[16]; | |
247 | ||
248 | /** | |
249 | * Motion vector cache. | |
250 | */ | |
251 | DECLARE_ALIGNED_8(int16_t, mv_cache[2][5*8][2]); | |
252 | DECLARE_ALIGNED_8(int8_t, ref_cache[2][5*8]); | |
253 | #define LIST_NOT_USED -1 //FIXME rename? | |
254 | #define PART_NOT_AVAILABLE -2 | |
255 | ||
256 | /** | |
257 | * is 1 if the specific list MV&references are set to 0,0,-2. | |
258 | */ | |
259 | int mv_cache_clean[2]; | |
260 | ||
261 | /** | |
262 | * number of neighbors (top and/or left) that used 8x8 dct | |
263 | */ | |
264 | int neighbor_transform_size; | |
265 | ||
266 | /** | |
267 | * block_offset[ 0..23] for frame macroblocks | |
268 | * block_offset[24..47] for field macroblocks | |
269 | */ | |
270 | int block_offset[2*(16+8)]; | |
271 | ||
272 | uint32_t *mb2b_xy; //FIXME are these 4 a good idea? | |
273 | uint32_t *mb2b8_xy; | |
274 | int b_stride; //FIXME use s->b4_stride | |
275 | int b8_stride; | |
276 | ||
277 | int mb_linesize; ///< may be equal to s->linesize or s->linesize*2, for mbaff | |
278 | int mb_uvlinesize; | |
279 | ||
280 | int emu_edge_width; | |
281 | int emu_edge_height; | |
282 | ||
283 | int halfpel_flag; | |
284 | int thirdpel_flag; | |
285 | ||
286 | int unknown_svq3_flag; | |
287 | int next_slice_index; | |
288 | ||
7978debd | 289 | SPS *sps_buffers[MAX_SPS_COUNT]; |
26b4fe82 AJ |
290 | SPS sps; ///< current sps |
291 | ||
7978debd | 292 | PPS *pps_buffers[MAX_PPS_COUNT]; |
26b4fe82 AJ |
293 | /** |
294 | * current pps | |
295 | */ | |
296 | PPS pps; //FIXME move to Picture perhaps? (->no) do we need that? | |
297 | ||
298 | uint32_t dequant4_buffer[6][52][16]; | |
299 | uint32_t dequant8_buffer[2][52][64]; | |
300 | uint32_t (*dequant4_coeff[6])[16]; | |
301 | uint32_t (*dequant8_coeff[2])[64]; | |
302 | int dequant_coeff_pps; ///< reinit tables when pps changes | |
303 | ||
304 | int slice_num; | |
b735aeea MN |
305 | uint16_t *slice_table_base; |
306 | uint16_t *slice_table; ///< slice_table_base + 2*mb_stride + 1 | |
26b4fe82 | 307 | int slice_type; |
e3e6f18f | 308 | int slice_type_nos; ///< S free slice type (SI/SP are remapped to I/P) |
26b4fe82 AJ |
309 | int slice_type_fixed; |
310 | ||
311 | //interlacing specific flags | |
312 | int mb_aff_frame; | |
313 | int mb_field_decoding_flag; | |
314 | int mb_mbaff; ///< mb_aff_frame && mb_field_decoding_flag | |
315 | ||
316 | unsigned int sub_mb_type[4]; | |
317 | ||
318 | //POC stuff | |
319 | int poc_lsb; | |
320 | int poc_msb; | |
321 | int delta_poc_bottom; | |
322 | int delta_poc[2]; | |
323 | int frame_num; | |
324 | int prev_poc_msb; ///< poc_msb of the last reference pic for POC type 0 | |
325 | int prev_poc_lsb; ///< poc_lsb of the last reference pic for POC type 0 | |
326 | int frame_num_offset; ///< for POC type 2 | |
327 | int prev_frame_num_offset; ///< for POC type 2 | |
328 | int prev_frame_num; ///< frame_num of the last pic for POC type 1/2 | |
329 | ||
330 | /** | |
f57e2af6 | 331 | * frame_num for frames or 2*frame_num+1 for field pics. |
26b4fe82 AJ |
332 | */ |
333 | int curr_pic_num; | |
334 | ||
335 | /** | |
336 | * max_frame_num or 2*max_frame_num for field pics. | |
337 | */ | |
338 | int max_pic_num; | |
339 | ||
340 | //Weighted pred stuff | |
341 | int use_weight; | |
342 | int use_weight_chroma; | |
343 | int luma_log2_weight_denom; | |
344 | int chroma_log2_weight_denom; | |
345 | int luma_weight[2][48]; | |
346 | int luma_offset[2][48]; | |
347 | int chroma_weight[2][48][2]; | |
348 | int chroma_offset[2][48][2]; | |
349 | int implicit_weight[48][48]; | |
350 | ||
351 | //deblock | |
352 | int deblocking_filter; ///< disable_deblocking_filter_idc with 1<->0 | |
353 | int slice_alpha_c0_offset; | |
354 | int slice_beta_offset; | |
355 | ||
356 | int redundant_pic_count; | |
357 | ||
358 | int direct_spatial_mv_pred; | |
359 | int dist_scale_factor[16]; | |
8b1fd554 | 360 | int dist_scale_factor_field[2][32]; |
f4d3382d MN |
361 | int map_col_to_list0[2][16+32]; |
362 | int map_col_to_list0_field[2][2][16+32]; | |
26b4fe82 AJ |
363 | |
364 | /** | |
365 | * num_ref_idx_l0/1_active_minus1 + 1 | |
366 | */ | |
367 | unsigned int ref_count[2]; ///< counts frames or fields, depending on current mb mode | |
368 | unsigned int list_count; | |
369 | Picture *short_ref[32]; | |
370 | Picture *long_ref[32]; | |
bcc3476c JD |
371 | Picture default_ref_list[2][32]; ///< base reference list for all slices of a coded picture |
372 | Picture ref_list[2][48]; /**< 0..15: frame refs, 16..47: mbaff field refs. | |
373 | Reordered version of default_ref_list | |
374 | according to picture reordering in slice header */ | |
b735aeea | 375 | int ref2frm[MAX_SLICES][2][64]; ///< reference to frame number lists, used in the loop filter, the first 2 are for -2,-1 |
64b9d48f | 376 | Picture *delayed_pic[MAX_DELAYED_PIC_COUNT+2]; //FIXME size? |
df8a7dff | 377 | int outputed_poc; |
26b4fe82 AJ |
378 | |
379 | /** | |
380 | * memory management control operations buffer. | |
381 | */ | |
382 | MMCO mmco[MAX_MMCO_COUNT]; | |
383 | int mmco_index; | |
384 | ||
385 | int long_ref_count; ///< number of actual long term references | |
386 | int short_ref_count; ///< number of actual short term references | |
387 | ||
388 | //data partitioning | |
389 | GetBitContext intra_gb; | |
390 | GetBitContext inter_gb; | |
391 | GetBitContext *intra_gb_ptr; | |
392 | GetBitContext *inter_gb_ptr; | |
393 | ||
ed5d7a53 | 394 | DECLARE_ALIGNED_16(DCTELEM, mb[16*24]); |
2cab6401 | 395 | DCTELEM mb_padding[256]; ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb |
26b4fe82 AJ |
396 | |
397 | /** | |
398 | * Cabac | |
399 | */ | |
400 | CABACContext cabac; | |
401 | uint8_t cabac_state[460]; | |
402 | int cabac_init_idc; | |
403 | ||
404 | /* 0x100 -> non null luma_dc, 0x80/0x40 -> non null chroma_dc (cb/cr), 0x?0 -> chroma_cbp(0,1,2), 0x0? luma_cbp */ | |
405 | uint16_t *cbp_table; | |
406 | int cbp; | |
407 | int top_cbp; | |
408 | int left_cbp; | |
409 | /* chroma_pred_mode for i4x4 or i16x16, else 0 */ | |
410 | uint8_t *chroma_pred_mode_table; | |
411 | int last_qscale_diff; | |
412 | int16_t (*mvd_table[2])[2]; | |
413 | DECLARE_ALIGNED_8(int16_t, mvd_cache[2][5*8][2]); | |
414 | uint8_t *direct_table; | |
415 | uint8_t direct_cache[5*8]; | |
416 | ||
417 | uint8_t zigzag_scan[16]; | |
418 | uint8_t zigzag_scan8x8[64]; | |
419 | uint8_t zigzag_scan8x8_cavlc[64]; | |
420 | uint8_t field_scan[16]; | |
421 | uint8_t field_scan8x8[64]; | |
422 | uint8_t field_scan8x8_cavlc[64]; | |
423 | const uint8_t *zigzag_scan_q0; | |
424 | const uint8_t *zigzag_scan8x8_q0; | |
425 | const uint8_t *zigzag_scan8x8_cavlc_q0; | |
426 | const uint8_t *field_scan_q0; | |
427 | const uint8_t *field_scan8x8_q0; | |
428 | const uint8_t *field_scan8x8_cavlc_q0; | |
429 | ||
430 | int x264_build; | |
afebe2f7 AÖ |
431 | |
432 | /** | |
433 | * @defgroup multithreading Members for slice based multithreading | |
434 | * @{ | |
435 | */ | |
436 | struct H264Context *thread_context[MAX_THREADS]; | |
437 | ||
438 | /** | |
439 | * current slice number, used to initalize slice_num of each thread/context | |
440 | */ | |
441 | int current_slice; | |
442 | ||
443 | /** | |
444 | * Max number of threads / contexts. | |
445 | * This is equal to AVCodecContext.thread_count unless | |
446 | * multithreaded decoding is impossible, in which case it is | |
447 | * reduced to 1. | |
448 | */ | |
449 | int max_contexts; | |
450 | ||
451 | /** | |
452 | * 1 if the single thread fallback warning has already been | |
453 | * displayed, 0 otherwise. | |
454 | */ | |
455 | int single_decode_warning; | |
456 | ||
457 | int last_slice_type; | |
458 | /** @} */ | |
459 | ||
64514ee8 AS |
460 | int mb_xy; |
461 | ||
1fdcdf8d | 462 | uint32_t svq3_watermark_key; |
26b4fe82 AJ |
463 | }H264Context; |
464 | ||
98790382 | 465 | #endif /* AVCODEC_H264_H */ |