FFmpeg coverage


Directory: ../../../ffmpeg/
File: src/libavcodec/mpegvideo.h
Date: 2022-01-16 20:33:26
Exec Total Coverage
Lines: 23 23 100.0%
Branches: 3 4 75.0%

Line Branch Exec Source
1 /*
2 * Generic DCT based hybrid video encoder
3 * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
4 * Copyright (c) 2002-2004 Michael Niedermayer
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 /**
24 * @file
25 * mpegvideo header.
26 */
27
28 #ifndef AVCODEC_MPEGVIDEO_H
29 #define AVCODEC_MPEGVIDEO_H
30
31 #include <float.h>
32
33 #include "avcodec.h"
34 #include "blockdsp.h"
35 #include "error_resilience.h"
36 #include "fdctdsp.h"
37 #include "get_bits.h"
38 #include "h264chroma.h"
39 #include "h263dsp.h"
40 #include "hpeldsp.h"
41 #include "idctdsp.h"
42 #include "internal.h"
43 #include "me_cmp.h"
44 #include "motion_est.h"
45 #include "mpegpicture.h"
46 #include "mpegvideodsp.h"
47 #include "mpegvideoencdsp.h"
48 #include "mpegvideodata.h"
49 #include "pixblockdsp.h"
50 #include "put_bits.h"
51 #include "ratecontrol.h"
52 #if FF_API_FLAG_TRUNCATED
53 #include "parser.h"
54 #endif
55 #include "mpegutils.h"
56 #include "qpeldsp.h"
57 #include "rl.h"
58 #include "videodsp.h"
59
60 #include "libavutil/opt.h"
61
62 #define MAX_THREADS 32
63
64 #define MAX_B_FRAMES 16
65
66 /**
67 * MpegEncContext.
68 */
69 typedef struct MpegEncContext {
70 AVClass *class;
71
72 int y_dc_scale, c_dc_scale;
73 int ac_pred;
74 int block_last_index[12]; ///< last non zero coefficient in block
75 int h263_aic; ///< Advanced INTRA Coding (AIC)
76
77 /* scantables */
78 ScanTable inter_scantable; ///< if inter == intra then intra should be used to reduce the cache usage
79 ScanTable intra_scantable;
80 ScanTable intra_h_scantable;
81 ScanTable intra_v_scantable;
82
83 /* WARNING: changes above this line require updates to hardcoded
84 * offsets used in ASM. */
85
86 struct AVCodecContext *avctx;
87 /* the following parameters must be initialized before encoding */
88 int width, height;///< picture size. must be a multiple of 16
89 int gop_size;
90 int intra_only; ///< if true, only intra pictures are generated
91 int64_t bit_rate; ///< wanted bit rate
92 enum OutputFormat out_format; ///< output format
93 int h263_pred; ///< use MPEG-4/H.263 ac/dc predictions
94 int pb_frame; ///< PB-frame mode (0 = none, 1 = base, 2 = improved)
95
96 /* the following codec id fields are deprecated in favor of codec_id */
97 int h263_plus; ///< H.263+ headers
98 int h263_flv; ///< use flv H.263 header
99
100 enum AVCodecID codec_id; /* see AV_CODEC_ID_xxx */
101 int fixed_qscale; ///< fixed qscale if non zero
102 int encoding; ///< true if we are encoding (vs decoding)
103 int max_b_frames; ///< max number of B-frames for encoding
104 int luma_elim_threshold;
105 int chroma_elim_threshold;
106 int strict_std_compliance; ///< strictly follow the std (MPEG-4, ...)
107 int workaround_bugs; ///< workaround bugs in encoders which cannot be detected automatically
108 int codec_tag; ///< internal codec_tag upper case converted from avctx codec_tag
109 /* the following fields are managed internally by the encoder */
110
111 /* sequence parameters */
112 int context_initialized;
113 int input_picture_number; ///< used to set pic->display_picture_number, should not be used for/by anything else
114 int coded_picture_number; ///< used to set pic->coded_picture_number, should not be used for/by anything else
115 int picture_number; //FIXME remove, unclear definition
116 int picture_in_gop_number; ///< 0-> first pic in gop, ...
117 int mb_width, mb_height; ///< number of MBs horizontally & vertically
118 int mb_stride; ///< mb_width+1 used for some arrays to allow simple addressing of left & top MBs without sig11
119 int b8_stride; ///< 2*mb_width+1 used for some 8x8 block arrays to allow simple addressing
120 int h_edge_pos, v_edge_pos;///< horizontal / vertical position of the right/bottom edge (pixel replication)
121 int mb_num; ///< number of MBs of a picture
122 ptrdiff_t linesize; ///< line size, in bytes, may be different from width
123 ptrdiff_t uvlinesize; ///< line size, for chroma in bytes, may be different from width
124 Picture *picture; ///< main picture buffer
125 Picture **input_picture; ///< next pictures on display order for encoding
126 Picture **reordered_input_picture; ///< pointer to the next pictures in coded order for encoding
127
128 int64_t user_specified_pts; ///< last non-zero pts from AVFrame which was passed into avcodec_send_frame()
129 /**
130 * pts difference between the first and second input frame, used for
131 * calculating dts of the first frame when there's a delay */
132 int64_t dts_delta;
133 /**
134 * reordered pts to be used as dts for the next output frame when there's
135 * a delay */
136 int64_t reordered_pts;
137
138 /** bit output */
139 PutBitContext pb;
140
141 int start_mb_y; ///< start mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
142 int end_mb_y; ///< end mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
143 struct MpegEncContext *thread_context[MAX_THREADS];
144 int slice_context_count; ///< number of used thread_contexts
145
146 /**
147 * copy of the previous picture structure.
148 * note, linesize & data, might not match the previous picture (for field pictures)
149 */
150 Picture last_picture;
151
152 /**
153 * copy of the next picture structure.
154 * note, linesize & data, might not match the next picture (for field pictures)
155 */
156 Picture next_picture;
157
158 /**
159 * copy of the source picture structure for encoding.
160 * note, linesize & data, might not match the source picture (for field pictures)
161 */
162 Picture new_picture;
163
164 /**
165 * copy of the current picture structure.
166 * note, linesize & data, might not match the current picture (for field pictures)
167 */
168 Picture current_picture; ///< buffer to store the decompressed current picture
169
170 Picture *last_picture_ptr; ///< pointer to the previous picture.
171 Picture *next_picture_ptr; ///< pointer to the next picture (for bidir pred)
172 Picture *current_picture_ptr; ///< pointer to the current picture
173 int last_dc[3]; ///< last DC values for MPEG-1
174 int16_t *dc_val_base;
175 int16_t *dc_val[3]; ///< used for MPEG-4 DC prediction, all 3 arrays must be continuous
176 const uint8_t *y_dc_scale_table; ///< qscale -> y_dc_scale table
177 const uint8_t *c_dc_scale_table; ///< qscale -> c_dc_scale table
178 const uint8_t *chroma_qscale_table; ///< qscale -> chroma_qscale (H.263)
179 uint8_t *coded_block_base;
180 uint8_t *coded_block; ///< used for coded block pattern prediction (msmpeg4v3, wmv1)
181 int16_t (*ac_val_base)[16];
182 int16_t (*ac_val[3])[16]; ///< used for MPEG-4 AC prediction, all 3 arrays must be continuous
183 int mb_skipped; ///< MUST BE SET only during DECODING
184 uint8_t *mbskip_table; /**< used to avoid copy if macroblock skipped (for black regions for example)
185 and used for B-frame encoding & decoding (contains skip table of next P-frame) */
186 uint8_t *mbintra_table; ///< used to avoid setting {ac, dc, cbp}-pred stuff to zero on inter MB decoding
187 uint8_t *cbp_table; ///< used to store cbp, ac_pred for partitioned decoding
188 uint8_t *pred_dir_table; ///< used to store pred_dir for partitioned decoding
189
190 ScratchpadContext sc;
191
192 int qscale; ///< QP
193 int chroma_qscale; ///< chroma QP
194 unsigned int lambda; ///< Lagrange multiplier used in rate distortion
195 unsigned int lambda2; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT
196 int *lambda_table;
197 int adaptive_quant; ///< use adaptive quantization
198 int dquant; ///< qscale difference to prev qscale
199 int pict_type; ///< AV_PICTURE_TYPE_I, AV_PICTURE_TYPE_P, AV_PICTURE_TYPE_B, ...
200 int vbv_delay;
201 int last_pict_type; //FIXME removes
202 int last_non_b_pict_type; ///< used for MPEG-4 gmc B-frames & ratecontrol
203 int droppable;
204 int frame_rate_index;
205 int last_lambda_for[5]; ///< last lambda for a specific pict type
206 int skipdct; ///< skip dct and code zero residual
207
208 /* motion compensation */
209 int unrestricted_mv; ///< mv can point outside of the coded picture
210 int h263_long_vectors; ///< use horrible H.263v1 long vector mode
211
212 BlockDSPContext bdsp;
213 FDCTDSPContext fdsp;
214 H264ChromaContext h264chroma;
215 HpelDSPContext hdsp;
216 IDCTDSPContext idsp;
217 MECmpContext mecc;
218 MpegVideoDSPContext mdsp;
219 MpegvideoEncDSPContext mpvencdsp;
220 PixblockDSPContext pdsp;
221 QpelDSPContext qdsp;
222 VideoDSPContext vdsp;
223 H263DSPContext h263dsp;
224 int f_code; ///< forward MV resolution
225 int b_code; ///< backward MV resolution for B-frames (MPEG-4)
226 int16_t (*p_mv_table_base)[2];
227 int16_t (*b_forw_mv_table_base)[2];
228 int16_t (*b_back_mv_table_base)[2];
229 int16_t (*b_bidir_forw_mv_table_base)[2];
230 int16_t (*b_bidir_back_mv_table_base)[2];
231 int16_t (*b_direct_mv_table_base)[2];
232 int16_t (*p_field_mv_table_base)[2];
233 int16_t (*b_field_mv_table_base)[2];
234 int16_t (*p_mv_table)[2]; ///< MV table (1MV per MB) P-frame encoding
235 int16_t (*b_forw_mv_table)[2]; ///< MV table (1MV per MB) forward mode B-frame encoding
236 int16_t (*b_back_mv_table)[2]; ///< MV table (1MV per MB) backward mode B-frame encoding
237 int16_t (*b_bidir_forw_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
238 int16_t (*b_bidir_back_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
239 int16_t (*b_direct_mv_table)[2]; ///< MV table (1MV per MB) direct mode B-frame encoding
240 int16_t (*p_field_mv_table[2][2])[2]; ///< MV table (2MV per MB) interlaced P-frame encoding
241 int16_t (*b_field_mv_table[2][2][2])[2];///< MV table (4MV per MB) interlaced B-frame encoding
242 uint8_t (*p_field_select_table[2]); ///< Only the first element is allocated
243 uint8_t (*b_field_select_table[2][2]); ///< Only the first element is allocated
244 int motion_est; ///< ME algorithm
245 int me_penalty_compensation;
246 int me_pre; ///< prepass for motion estimation
247 int mv_dir;
248 #define MV_DIR_FORWARD 1
249 #define MV_DIR_BACKWARD 2
250 #define MV_DIRECT 4 ///< bidirectional mode where the difference equals the MV of the last P/S/I-Frame (MPEG-4)
251 int mv_type;
252 #define MV_TYPE_16X16 0 ///< 1 vector for the whole mb
253 #define MV_TYPE_8X8 1 ///< 4 vectors (H.263, MPEG-4 4MV)
254 #define MV_TYPE_16X8 2 ///< 2 vectors, one per 16x8 block
255 #define MV_TYPE_FIELD 3 ///< 2 vectors, one per field
256 #define MV_TYPE_DMV 4 ///< 2 vectors, special mpeg2 Dual Prime Vectors
257 /**motion vectors for a macroblock
258 first coordinate : 0 = forward 1 = backward
259 second " : depend on type
260 third " : 0 = x, 1 = y
261 */
262 int mv[2][4][2];
263 int field_select[2][2];
264 int last_mv[2][2][2]; ///< last MV, used for MV prediction in MPEG-1 & B-frame MPEG-4
265 const uint8_t *fcode_tab; ///< smallest fcode needed for each MV
266 int16_t direct_scale_mv[2][64]; ///< precomputed to avoid divisions in ff_mpeg4_set_direct_mv
267
268 MotionEstContext me;
269
270 int no_rounding; /**< apply no rounding to motion compensation (MPEG-4, msmpeg4, ...)
271 for B-frames rounding mode is always 0 */
272
273 /* macroblock layer */
274 int mb_x, mb_y;
275 int mb_skip_run;
276 int mb_intra;
277 uint16_t *mb_type; ///< Table for candidate MB types for encoding (defines in mpegutils.h)
278
279 int block_index[6]; ///< index to current MB in block based arrays with edges
280 int block_wrap[6];
281 uint8_t *dest[3];
282
283 int *mb_index2xy; ///< mb_index -> mb_x + mb_y*mb_stride
284
285 /** matrix transmitted in the bitstream */
286 uint16_t intra_matrix[64];
287 uint16_t chroma_intra_matrix[64];
288 uint16_t inter_matrix[64];
289 uint16_t chroma_inter_matrix[64];
290
291 int intra_quant_bias; ///< bias for the quantizer
292 int inter_quant_bias; ///< bias for the quantizer
293 int min_qcoeff; ///< minimum encodable coefficient
294 int max_qcoeff; ///< maximum encodable coefficient
295 int ac_esc_length; ///< num of bits needed to encode the longest esc
296 uint8_t *intra_ac_vlc_length;
297 uint8_t *intra_ac_vlc_last_length;
298 uint8_t *intra_chroma_ac_vlc_length;
299 uint8_t *intra_chroma_ac_vlc_last_length;
300 uint8_t *inter_ac_vlc_length;
301 uint8_t *inter_ac_vlc_last_length;
302 uint8_t *luma_dc_vlc_length;
303 #define UNI_AC_ENC_INDEX(run,level) ((run)*128 + (level))
304
305 int coded_score[12];
306
307 /** precomputed matrix (combine qscale and DCT renorm) */
308 int (*q_intra_matrix)[64];
309 int (*q_chroma_intra_matrix)[64];
310 int (*q_inter_matrix)[64];
311 /** identical to the above but for MMX & these are not permutated, second 64 entries are bias*/
312 uint16_t (*q_intra_matrix16)[2][64];
313 uint16_t (*q_chroma_intra_matrix16)[2][64];
314 uint16_t (*q_inter_matrix16)[2][64];
315
316 /* noise reduction */
317 int (*dct_error_sum)[64];
318 int dct_count[2];
319 uint16_t (*dct_offset)[64];
320
321 /* bit rate control */
322 int64_t total_bits;
323 int frame_bits; ///< bits used for the current frame
324 int stuffing_bits; ///< bits used for stuffing
325 int next_lambda; ///< next lambda used for retrying to encode a frame
326 RateControlContext rc_context; ///< contains stuff only accessed in ratecontrol.c
327
328 /* statistics, used for 2-pass encoding */
329 int mv_bits;
330 int header_bits;
331 int i_tex_bits;
332 int p_tex_bits;
333 int i_count;
334 int f_count;
335 int b_count;
336 int skip_count;
337 int misc_bits; ///< cbp, mb_type
338 int last_bits; ///< temp var used for calculating the above vars
339
340 /* error concealment / resync */
341 int resync_mb_x; ///< x position of last resync marker
342 int resync_mb_y; ///< y position of last resync marker
343 GetBitContext last_resync_gb; ///< used to search for the next resync marker
344 int mb_num_left; ///< number of MBs left in this video packet (for partitioned Slices only)
345
346 #if FF_API_FLAG_TRUNCATED
347 ParseContext parse_context;
348 #endif
349
350 /* H.263 specific */
351 int gob_index;
352 int obmc; ///< overlapped block motion compensation
353 int mb_info; ///< interval for outputting info about mb offsets as side data
354 int prev_mb_info, last_mb_info;
355 uint8_t *mb_info_ptr;
356 int mb_info_size;
357 int ehc_mode;
358
359 /* H.263+ specific */
360 int umvplus; ///< == H.263+ && unrestricted_mv
361 int h263_aic_dir; ///< AIC direction: 0 = left, 1 = top
362 int h263_slice_structured;
363 int alt_inter_vlc; ///< alternative inter vlc
364 int modified_quant;
365 int loop_filter;
366 int custom_pcf;
367
368 /* MPEG-4 specific */
369 int studio_profile;
370 int dct_precision;
371 ///< number of bits to represent the fractional part of time (encoder only)
372 int time_increment_bits;
373 int last_time_base;
374 int time_base; ///< time in seconds of last I,P,S Frame
375 int64_t time; ///< time of current frame
376 int64_t last_non_b_time;
377 uint16_t pp_time; ///< time distance between the last 2 p,s,i frames
378 uint16_t pb_time; ///< time distance between the last b and p,s,i frame
379 uint16_t pp_field_time;
380 uint16_t pb_field_time; ///< like above, just for interlaced
381 int real_sprite_warping_points;
382 int sprite_offset[2][2]; ///< sprite offset[isChroma][isMVY]
383 int sprite_delta[2][2]; ///< sprite_delta [isY][isMVY]
384 int mcsel;
385 int quant_precision;
386 int quarter_sample; ///< 1->qpel, 0->half pel ME/MC
387 int aspect_ratio_info; //FIXME remove
388 int sprite_warping_accuracy;
389 int data_partitioning; ///< data partitioning flag from header
390 int partitioned_frame; ///< is current frame partitioned
391 int low_delay; ///< no reordering needed / has no B-frames
392 int vo_type;
393 PutBitContext tex_pb; ///< used for data partitioned VOPs
394 PutBitContext pb2; ///< used for data partitioned VOPs
395 int mpeg_quant;
396 int padding_bug_score; ///< used to detect the VERY common padding bug in MPEG-4
397
398 /* divx specific, used to workaround (many) bugs in divx5 */
399 int divx_packed;
400 uint8_t *bitstream_buffer; //Divx 5.01 puts several frames in a single one, this is used to reorder them
401 int bitstream_buffer_size;
402 unsigned int allocated_bitstream_buffer_size;
403
404 /* RV10 specific */
405 int rv10_version; ///< RV10 version: 0 or 3
406 int rv10_first_dc_coded[3];
407
408 /* MJPEG specific */
409 struct MJpegContext *mjpeg_ctx;
410 int esc_pos;
411
412 /* MSMPEG4 specific */
413 int mv_table_index;
414 int rl_table_index;
415 int rl_chroma_table_index;
416 int dc_table_index;
417 int use_skip_mb_code;
418 int slice_height; ///< in macroblocks
419 int first_slice_line; ///< used in MPEG-4 too to handle resync markers
420 int flipflop_rounding;
421 int msmpeg4_version; ///< 0=not msmpeg4, 1=mp41, 2=mp42, 3=mp43/divx3 4=wmv1/7 5=wmv2/8
422 int per_mb_rl_table;
423 int esc3_level_length;
424 int esc3_run_length;
425 /** [mb_intra][isChroma][level][run][last] */
426 int (*ac_stats)[2][MAX_LEVEL+1][MAX_RUN+1][2];
427 int inter_intra_pred;
428 int mspel;
429
430 /* SpeedHQ specific */
431 int slice_start;
432
433 /* decompression specific */
434 GetBitContext gb;
435
436 /* MPEG-1 specific */
437 int gop_picture_number; ///< index of the first picture of a GOP based on fake_pic_num & MPEG-1 specific
438 int last_mv_dir; ///< last mv_dir, used for B-frame encoding
439 uint8_t *vbv_delay_ptr; ///< pointer to vbv_delay in the bitstream
440
441 /* MPEG-2-specific - I wished not to have to support this mess. */
442 int progressive_sequence;
443 int mpeg_f_code[2][2];
444
445 // picture structure defines are loaded from mpegutils.h
446 int picture_structure;
447
448 int64_t timecode_frame_start; ///< GOP timecode frame start number, in non drop frame format
449 int intra_dc_precision;
450 int frame_pred_frame_dct;
451 int top_field_first;
452 int concealment_motion_vectors;
453 int q_scale_type;
454 int brd_scale;
455 int intra_vlc_format;
456 int alternate_scan;
457 #define VIDEO_FORMAT_COMPONENT 0
458 #define VIDEO_FORMAT_PAL 1
459 #define VIDEO_FORMAT_NTSC 2
460 #define VIDEO_FORMAT_SECAM 3
461 #define VIDEO_FORMAT_MAC 4
462 #define VIDEO_FORMAT_UNSPECIFIED 5
463 int repeat_first_field;
464 int chroma_420_type;
465 int chroma_format;
466 #define CHROMA_420 1
467 #define CHROMA_422 2
468 #define CHROMA_444 3
469 int chroma_x_shift;//depend on pix_format, that depend on chroma_format
470 int chroma_y_shift;
471
472 int progressive_frame;
473 int full_pel[2];
474 int interlaced_dct;
475 int first_field; ///< is 1 for the first field of a field picture 0 otherwise
476
477 /* RTP specific */
478 int rtp_mode;
479 int rtp_payload_size;
480
481 uint8_t *ptr_lastgob;
482 int swap_uv; //vcr2 codec is an MPEG-2 variant with U and V swapped
483 int pack_pblocks; //xvmc needs to keep blocks without gaps.
484 int16_t (*pblocks[12])[64];
485
486 int16_t (*block)[64]; ///< points to one of the following blocks
487 int16_t (*blocks)[12][64]; // for HQ mode we need to keep the best block
488 int (*decode_mb)(struct MpegEncContext *s, int16_t block[12][64]); // used by some codecs to avoid a switch()
489
490 int32_t (*block32)[12][64];
491 int dpcm_direction; // 0 = DCT, 1 = DPCM top to bottom scan, -1 = DPCM bottom to top scan
492 int16_t (*dpcm_macroblock)[3][256];
493
494 #define SLICE_OK 0
495 #define SLICE_ERROR -1
496 #define SLICE_END -2 ///<end marker found
497 #define SLICE_NOEND -3 ///<no end marker or error found but mb count exceeded
498
499 void (*dct_unquantize_mpeg1_intra)(struct MpegEncContext *s,
500 int16_t *block/*align 16*/, int n, int qscale);
501 void (*dct_unquantize_mpeg1_inter)(struct MpegEncContext *s,
502 int16_t *block/*align 16*/, int n, int qscale);
503 void (*dct_unquantize_mpeg2_intra)(struct MpegEncContext *s,
504 int16_t *block/*align 16*/, int n, int qscale);
505 void (*dct_unquantize_mpeg2_inter)(struct MpegEncContext *s,
506 int16_t *block/*align 16*/, int n, int qscale);
507 void (*dct_unquantize_h263_intra)(struct MpegEncContext *s,
508 int16_t *block/*align 16*/, int n, int qscale);
509 void (*dct_unquantize_h263_inter)(struct MpegEncContext *s,
510 int16_t *block/*align 16*/, int n, int qscale);
511 void (*dct_unquantize_intra)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
512 int16_t *block/*align 16*/, int n, int qscale);
513 void (*dct_unquantize_inter)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
514 int16_t *block/*align 16*/, int n, int qscale);
515 int (*dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
516 int (*fast_dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
517 void (*denoise_dct)(struct MpegEncContext *s, int16_t *block);
518
519 int mpv_flags; ///< flags set by private options
520 int quantizer_noise_shaping;
521
522 /**
523 * ratecontrol qmin qmax limiting method
524 * 0-> clipping, 1-> use a nice continuous function to limit qscale within qmin/qmax.
525 */
526 float rc_qsquish;
527 float rc_qmod_amp;
528 int rc_qmod_freq;
529 float rc_initial_cplx;
530 float rc_buffer_aggressivity;
531 float border_masking;
532 int lmin, lmax;
533 int vbv_ignore_qmax;
534
535 char *rc_eq;
536
537 /* temp buffers for rate control */
538 float *cplx_tab, *bits_tab;
539
540 /* flag to indicate a reinitialization is required, e.g. after
541 * a frame size change */
542 int context_reinit;
543
544 ERContext er;
545
546 int error_rate;
547
548 /* temporary frames used by b_frame_strategy = 2 */
549 AVFrame *tmp_frames[MAX_B_FRAMES + 2];
550 int b_frame_strategy;
551 int b_sensitivity;
552
553 /* frame skip options for encoding */
554 int frame_skip_threshold;
555 int frame_skip_factor;
556 int frame_skip_exp;
557 int frame_skip_cmp;
558
559 int scenechange_threshold;
560 int noise_reduction;
561
562 int intra_penalty;
563
564 #if FF_API_MPEGVIDEO_OPTS || FF_API_MJPEG_PRED
565 int dummy; ///< used as target for deprecated options
566 #endif
567 } MpegEncContext;
568
569 /* mpegvideo_enc common options */
570 #define FF_MPV_FLAG_SKIP_RD 0x0001
571 #define FF_MPV_FLAG_STRICT_GOP 0x0002
572 #define FF_MPV_FLAG_QP_RD 0x0004
573 #define FF_MPV_FLAG_CBP_RD 0x0008
574 #define FF_MPV_FLAG_NAQ 0x0010
575 #define FF_MPV_FLAG_MV0 0x0020
576
577 #define FF_MPV_OPT_CMP_FUNC \
578 { "sad", "Sum of absolute differences, fast", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
579 { "sse", "Sum of squared errors", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
580 { "satd", "Sum of absolute Hadamard transformed differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SATD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
581 { "dct", "Sum of absolute DCT transformed differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCT }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
582 { "psnr", "Sum of squared quantization errors, low quality", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_PSNR }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
583 { "bit", "Number of bits needed for the block", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_BIT }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
584 { "rd", "Rate distortion optimal, slow", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_RD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
585 { "zero", "Zero", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_ZERO }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
586 { "vsad", "Sum of absolute vertical differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_VSAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
587 { "vsse", "Sum of squared vertical differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_VSSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
588 { "nsse", "Noise preserving sum of squared differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_NSSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
589 { "dct264", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCT264 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
590 { "dctmax", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCTMAX }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
591 { "chroma", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_CHROMA }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
592 { "msad", "Sum of absolute differences, median predicted", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_MEDIAN_SAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }
593
594 #ifndef FF_MPV_OFFSET
595 #define FF_MPV_OFFSET(x) offsetof(MpegEncContext, x)
596 #endif
597 #define FF_MPV_OPT_FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
598 #define FF_MPV_COMMON_OPTS \
599 FF_MPV_OPT_CMP_FUNC, \
600 { "mpv_flags", "Flags common for all mpegvideo-based encoders.", FF_MPV_OFFSET(mpv_flags), AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "mpv_flags" },\
601 { "skip_rd", "RD optimal MB level residual skipping", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_SKIP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
602 { "strict_gop", "Strictly enforce gop size", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_STRICT_GOP }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
603 { "qp_rd", "Use rate distortion optimization for qp selection", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_QP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
604 { "cbp_rd", "use rate distortion optimization for CBP", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_CBP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
605 { "naq", "normalize adaptive quantization", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_NAQ }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
606 { "mv0", "always try a mb with mv=<0,0>", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_MV0 }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
607 { "luma_elim_threshold", "single coefficient elimination threshold for luminance (negative values also consider dc coefficient)",\
608 FF_MPV_OFFSET(luma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
609 { "chroma_elim_threshold", "single coefficient elimination threshold for chrominance (negative values also consider dc coefficient)",\
610 FF_MPV_OFFSET(chroma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
611 { "quantizer_noise_shaping", NULL, FF_MPV_OFFSET(quantizer_noise_shaping), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
612 { "error_rate", "Simulate errors in the bitstream to test error concealment.", \
613 FF_MPV_OFFSET(error_rate), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
614 {"qsquish", "how to keep quantizer between qmin and qmax (0 = clip, 1 = use differentiable function)", \
615 FF_MPV_OFFSET(rc_qsquish), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, 0, 99, FF_MPV_OPT_FLAGS}, \
616 {"rc_qmod_amp", "experimental quantizer modulation", FF_MPV_OFFSET(rc_qmod_amp), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
617 {"rc_qmod_freq", "experimental quantizer modulation", FF_MPV_OFFSET(rc_qmod_freq), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS}, \
618 {"rc_eq", "Set rate control equation. When computing the expression, besides the standard functions " \
619 "defined in the section 'Expression Evaluation', the following functions are available: " \
620 "bits2qp(bits), qp2bits(qp). Also the following constants are available: iTex pTex tex mv " \
621 "fCode iCount mcVar var isI isP isB avgQP qComp avgIITex avgPITex avgPPTex avgBPTex avgTex.", \
622 FF_MPV_OFFSET(rc_eq), AV_OPT_TYPE_STRING, .flags = FF_MPV_OPT_FLAGS }, \
623 {"rc_init_cplx", "initial complexity for 1-pass encoding", FF_MPV_OFFSET(rc_initial_cplx), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
624 {"rc_buf_aggressivity", "currently useless", FF_MPV_OFFSET(rc_buffer_aggressivity), AV_OPT_TYPE_FLOAT, {.dbl = 1.0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
625 {"border_mask", "increase the quantizer for macroblocks close to borders", FF_MPV_OFFSET(border_masking), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
626 {"lmin", "minimum Lagrange factor (VBR)", FF_MPV_OFFSET(lmin), AV_OPT_TYPE_INT, {.i64 = 2*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
627 {"lmax", "maximum Lagrange factor (VBR)", FF_MPV_OFFSET(lmax), AV_OPT_TYPE_INT, {.i64 = 31*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
628 {"ibias", "intra quant bias", FF_MPV_OFFSET(intra_quant_bias), AV_OPT_TYPE_INT, {.i64 = FF_DEFAULT_QUANT_BIAS }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
629 {"pbias", "inter quant bias", FF_MPV_OFFSET(inter_quant_bias), AV_OPT_TYPE_INT, {.i64 = FF_DEFAULT_QUANT_BIAS }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
630 {"motion_est", "motion estimation algorithm", FF_MPV_OFFSET(motion_est), AV_OPT_TYPE_INT, {.i64 = FF_ME_EPZS }, FF_ME_ZERO, FF_ME_XONE, FF_MPV_OPT_FLAGS, "motion_est" }, \
631 { "zero", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_ZERO }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
632 { "epzs", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_EPZS }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
633 { "xone", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_XONE }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
634 {"skip_threshold", "Frame skip threshold", FF_MPV_OFFSET(frame_skip_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
635 {"skip_factor", "Frame skip factor", FF_MPV_OFFSET(frame_skip_factor), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
636 {"skip_exp", "Frame skip exponent", FF_MPV_OFFSET(frame_skip_exp), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
637 {"skip_cmp", "Frame skip compare function", FF_MPV_OFFSET(frame_skip_cmp), AV_OPT_TYPE_INT, {.i64 = FF_CMP_DCTMAX }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
638 {"sc_threshold", "Scene change threshold", FF_MPV_OFFSET(scenechange_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
639 {"noise_reduction", "Noise reduction", FF_MPV_OFFSET(noise_reduction), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
640 {"ps", "RTP payload size in bytes", FF_MPV_OFFSET(rtp_payload_size), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
641 {"mepc", "Motion estimation bitrate penalty compensation (1.0 = 256)", FF_MPV_OFFSET(me_penalty_compensation), AV_OPT_TYPE_INT, {.i64 = 256 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
642 {"mepre", "pre motion estimation", FF_MPV_OFFSET(me_pre), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
643 {"intra_penalty", "Penalty for intra blocks in block decision", FF_MPV_OFFSET(intra_penalty), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, INT_MAX/2, FF_MPV_OPT_FLAGS }, \
644
645 #define FF_MPV_COMMON_BFRAME_OPTS \
646 {"b_strategy", "Strategy to choose between I/P/B-frames", FF_MPV_OFFSET(b_frame_strategy), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 2, FF_MPV_OPT_FLAGS }, \
647 {"b_sensitivity", "Adjust sensitivity of b_frame_strategy 1", FF_MPV_OFFSET(b_sensitivity), AV_OPT_TYPE_INT, {.i64 = 40 }, 1, INT_MAX, FF_MPV_OPT_FLAGS }, \
648 {"brd_scale", "Downscale frames for dynamic B-frame decision", FF_MPV_OFFSET(brd_scale), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 3, FF_MPV_OPT_FLAGS },
649
650 #if FF_API_MPEGVIDEO_OPTS
651 #define FF_MPV_DEPRECATED_MPEG_QUANT_OPT \
652 { "mpeg_quant", "Deprecated, does nothing", FF_MPV_OFFSET(mpeg_quant), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 0, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
653 #define FF_MPV_DEPRECATED_A53_CC_OPT \
654 { "a53cc", "Deprecated, does nothing", FF_MPV_OFFSET(dummy), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
655 #define FF_MPV_DEPRECATED_MATRIX_OPT \
656 { "force_duplicated_matrix", "Deprecated, does nothing", FF_MPV_OFFSET(dummy), AV_OPT_TYPE_BOOL, {.i64 = 0 }, 0, 1, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
657 #define FF_MPV_DEPRECATED_BFRAME_OPTS \
658 { "b_strategy", "Deprecated, does nothing", FF_MPV_OFFSET(b_frame_strategy), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 2, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED }, \
659 { "b_sensitivity", "Deprecated, does nothing", FF_MPV_OFFSET(b_sensitivity), AV_OPT_TYPE_INT, { .i64 = 40 }, 1, INT_MAX, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED }, \
660 { "brd_scale", "Deprecated, does nothing", FF_MPV_OFFSET(brd_scale), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 3, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
661 #endif
662
663 extern const AVClass ff_mpv_enc_class;
664
665 /**
666 * Set the given MpegEncContext to common defaults (same for encoding
667 * and decoding). The changed fields will not depend upon the prior
668 * state of the MpegEncContext.
669 */
670 void ff_mpv_common_defaults(MpegEncContext *s);
671
672 void ff_dct_encode_init_x86(MpegEncContext *s);
673
674 int ff_mpv_common_init(MpegEncContext *s);
675 void ff_mpv_common_init_arm(MpegEncContext *s);
676 void ff_mpv_common_init_axp(MpegEncContext *s);
677 void ff_mpv_common_init_neon(MpegEncContext *s);
678 void ff_mpv_common_init_ppc(MpegEncContext *s);
679 void ff_mpv_common_init_x86(MpegEncContext *s);
680 void ff_mpv_common_init_mips(MpegEncContext *s);
681 /**
682 * Initialize an MpegEncContext's thread contexts. Presumes that
683 * slice_context_count is already set and that all the fields
684 * that are freed/reset in free_duplicate_context() are NULL.
685 */
686 int ff_mpv_init_duplicate_contexts(MpegEncContext *s);
687 /**
688 * Initialize and allocates MpegEncContext fields dependent on the resolution.
689 */
690 int ff_mpv_init_context_frame(MpegEncContext *s);
691 /**
692 * Frees and resets MpegEncContext fields depending on the resolution
693 * as well as the slice thread contexts.
694 * Is used during resolution changes to avoid a full reinitialization of the
695 * codec.
696 */
697 void ff_mpv_free_context_frame(MpegEncContext *s);
698
699 int ff_mpv_common_frame_size_change(MpegEncContext *s);
700 void ff_mpv_common_end(MpegEncContext *s);
701
702 /**
703 * Initialize the given MpegEncContext for decoding.
704 * the changed fields will not depend upon
705 * the prior state of the MpegEncContext.
706 */
707 void ff_mpv_decode_init(MpegEncContext *s, AVCodecContext *avctx);
708 void ff_mpv_reconstruct_mb(MpegEncContext *s, int16_t block[12][64]);
709 void ff_mpv_report_decode_progress(MpegEncContext *s);
710
711 int ff_mpv_frame_start(MpegEncContext *s, AVCodecContext *avctx);
712 void ff_mpv_frame_end(MpegEncContext *s);
713
714 int ff_mpv_encode_init(AVCodecContext *avctx);
715 void ff_mpv_encode_init_x86(MpegEncContext *s);
716
717 int ff_mpv_encode_end(AVCodecContext *avctx);
718 int ff_mpv_encode_picture(AVCodecContext *avctx, AVPacket *pkt,
719 const AVFrame *frame, int *got_packet);
720 int ff_mpv_reallocate_putbitbuffer(MpegEncContext *s, size_t threshold, size_t size_increase);
721
722 void ff_clean_intra_table_entries(MpegEncContext *s);
723 void ff_mpeg_draw_horiz_band(MpegEncContext *s, int y, int h);
724 void ff_mpeg_flush(AVCodecContext *avctx);
725
726 void ff_print_debug_info(MpegEncContext *s, Picture *p, AVFrame *pict);
727
728 int ff_mpv_export_qp_table(MpegEncContext *s, AVFrame *f, Picture *p, int qp_type);
729
730 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix);
731
732 int ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src);
733 int ff_mpeg_update_thread_context(AVCodecContext *dst, const AVCodecContext *src);
734 void ff_set_qscale(MpegEncContext * s, int qscale);
735
736 void ff_mpv_idct_init(MpegEncContext *s);
737 int ff_dct_encode_init(MpegEncContext *s);
738 void ff_convert_matrix(MpegEncContext *s, int (*qmat)[64], uint16_t (*qmat16)[2][64],
739 const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra);
740 int ff_dct_quantize_c(MpegEncContext *s, int16_t *block, int n, int qscale, int *overflow);
741 void ff_block_permute(int16_t *block, uint8_t *permutation,
742 const uint8_t *scantable, int last);
743 void ff_init_block_index(MpegEncContext *s);
744
745 void ff_mpv_motion(MpegEncContext *s,
746 uint8_t *dest_y, uint8_t *dest_cb,
747 uint8_t *dest_cr, int dir,
748 uint8_t **ref_picture,
749 op_pixels_func (*pix_op)[4],
750 qpel_mc_func (*qpix_op)[16]);
751
752 6656257 static inline void ff_update_block_index(MpegEncContext *s){
753
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 6656257 times.
6656257 const int bytes_per_pixel = 1 + (s->avctx->bits_per_raw_sample > 8);
754 6656257 const int block_size= (8*bytes_per_pixel) >> s->avctx->lowres;
755
756 6656257 s->block_index[0]+=2;
757 6656257 s->block_index[1]+=2;
758 6656257 s->block_index[2]+=2;
759 6656257 s->block_index[3]+=2;
760 6656257 s->block_index[4]++;
761 6656257 s->block_index[5]++;
762 6656257 s->dest[0]+= 2*block_size;
763 6656257 s->dest[1]+= (2 >> s->chroma_x_shift) * block_size;
764 6656257 s->dest[2]+= (2 >> s->chroma_x_shift) * block_size;
765 6656257 }
766
767 4992717 static inline int get_bits_diff(MpegEncContext *s){
768 4992717 const int bits= put_bits_count(&s->pb);
769 4992717 const int last= s->last_bits;
770
771 4992717 s->last_bits = bits;
772
773 4992717 return bits - last;
774 }
775
776 271837 static inline int mpeg_get_qscale(MpegEncContext *s)
777 {
778 271837 int qscale = get_bits(&s->gb, 5);
779
2/2
✓ Branch 0 taken 210729 times.
✓ Branch 1 taken 61108 times.
271837 if (s->q_scale_type)
780 210729 return ff_mpeg2_non_linear_qscale[qscale];
781 else
782 61108 return qscale << 1;
783 }
784
785 #endif /* AVCODEC_MPEGVIDEO_H */
786