libx264: Make sure the extradata are padded
[libav.git] / libavcodec / libx264.c
1 /*
2 * H.264 encoding using the x264 library
3 * Copyright (C) 2005 Mans Rullgard <mans@mansr.com>
4 *
5 * This file is part of Libav.
6 *
7 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 #include "libavutil/internal.h"
23 #include "libavutil/opt.h"
24 #include "libavutil/mem.h"
25 #include "libavutil/pixdesc.h"
26 #include "libavutil/stereo3d.h"
27 #include "avcodec.h"
28 #include "internal.h"
29
30 #if defined(_MSC_VER)
31 #define X264_API_IMPORTS 1
32 #endif
33
34 #include <x264.h>
35 #include <float.h>
36 #include <math.h>
37 #include <stdio.h>
38 #include <stdlib.h>
39 #include <string.h>
40
41 typedef struct X264Context {
42 AVClass *class;
43 x264_param_t params;
44 x264_t *enc;
45 x264_picture_t pic;
46 uint8_t *sei;
47 int sei_size;
48 char *preset;
49 char *tune;
50 char *profile;
51 int fastfirstpass;
52 float crf;
53 float crf_max;
54 int cqp;
55 int aq_mode;
56 float aq_strength;
57 char *psy_rd;
58 int psy;
59 int rc_lookahead;
60 int weightp;
61 int weightb;
62 int ssim;
63 int intra_refresh;
64 int bluray_compat;
65 int b_bias;
66 int b_pyramid;
67 int mixed_refs;
68 int dct8x8;
69 int fast_pskip;
70 int aud;
71 int mbtree;
72 char *deblock;
73 float cplxblur;
74 char *partitions;
75 int direct_pred;
76 int slice_max_size;
77 char *stats;
78 int nal_hrd;
79 int motion_est;
80 int forced_idr;
81 char *x264_params;
82 } X264Context;
83
84 static void X264_log(void *p, int level, const char *fmt, va_list args)
85 {
86 static const int level_map[] = {
87 [X264_LOG_ERROR] = AV_LOG_ERROR,
88 [X264_LOG_WARNING] = AV_LOG_WARNING,
89 [X264_LOG_INFO] = AV_LOG_INFO,
90 [X264_LOG_DEBUG] = AV_LOG_DEBUG
91 };
92
93 if (level < 0 || level > X264_LOG_DEBUG)
94 return;
95
96 av_vlog(p, level_map[level], fmt, args);
97 }
98
99
100 static int encode_nals(AVCodecContext *ctx, AVPacket *pkt,
101 x264_nal_t *nals, int nnal)
102 {
103 X264Context *x4 = ctx->priv_data;
104 uint8_t *p;
105 int i, size = x4->sei_size, ret;
106
107 if (!nnal)
108 return 0;
109
110 for (i = 0; i < nnal; i++)
111 size += nals[i].i_payload;
112
113 if ((ret = ff_alloc_packet(pkt, size)) < 0)
114 return ret;
115
116 p = pkt->data;
117
118 /* Write the SEI as part of the first frame. */
119 if (x4->sei_size > 0 && nnal > 0) {
120 memcpy(p, x4->sei, x4->sei_size);
121 p += x4->sei_size;
122 x4->sei_size = 0;
123 }
124
125 for (i = 0; i < nnal; i++){
126 memcpy(p, nals[i].p_payload, nals[i].i_payload);
127 p += nals[i].i_payload;
128 }
129
130 return 1;
131 }
132
133 static void reconfig_encoder(AVCodecContext *ctx, const AVFrame *frame)
134 {
135 X264Context *x4 = ctx->priv_data;
136 AVFrameSideData *side_data;
137
138
139 if (x4->params.b_tff != frame->top_field_first) {
140 x4->params.b_tff = frame->top_field_first;
141 x264_encoder_reconfig(x4->enc, &x4->params);
142 }
143 if (x4->params.vui.i_sar_height != ctx->sample_aspect_ratio.den ||
144 x4->params.vui.i_sar_width != ctx->sample_aspect_ratio.num) {
145 x4->params.vui.i_sar_height = ctx->sample_aspect_ratio.den;
146 x4->params.vui.i_sar_width = ctx->sample_aspect_ratio.num;
147 x264_encoder_reconfig(x4->enc, &x4->params);
148 }
149
150 if (x4->params.rc.i_vbv_buffer_size != ctx->rc_buffer_size / 1000 ||
151 x4->params.rc.i_vbv_max_bitrate != ctx->rc_max_rate / 1000) {
152 x4->params.rc.i_vbv_buffer_size = ctx->rc_buffer_size / 1000;
153 x4->params.rc.i_vbv_max_bitrate = ctx->rc_max_rate / 1000;
154 x264_encoder_reconfig(x4->enc, &x4->params);
155 }
156
157 if (x4->params.rc.i_rc_method == X264_RC_ABR &&
158 x4->params.rc.i_bitrate != ctx->bit_rate / 1000) {
159 x4->params.rc.i_bitrate = ctx->bit_rate / 1000;
160 x264_encoder_reconfig(x4->enc, &x4->params);
161 }
162
163 if (x4->crf >= 0 &&
164 x4->params.rc.i_rc_method == X264_RC_CRF &&
165 x4->params.rc.f_rf_constant != x4->crf) {
166 x4->params.rc.f_rf_constant = x4->crf;
167 x264_encoder_reconfig(x4->enc, &x4->params);
168 }
169
170 if (x4->params.rc.i_rc_method == X264_RC_CQP &&
171 x4->params.rc.i_qp_constant != x4->cqp) {
172 x4->params.rc.i_qp_constant = x4->cqp;
173 x264_encoder_reconfig(x4->enc, &x4->params);
174 }
175
176 if (x4->crf_max >= 0 &&
177 x4->params.rc.f_rf_constant_max != x4->crf_max) {
178 x4->params.rc.f_rf_constant_max = x4->crf_max;
179 x264_encoder_reconfig(x4->enc, &x4->params);
180 }
181
182 side_data = av_frame_get_side_data(frame, AV_FRAME_DATA_STEREO3D);
183 if (side_data) {
184 AVStereo3D *stereo = (AVStereo3D *)side_data->data;
185 int fpa_type;
186
187 switch (stereo->type) {
188 case AV_STEREO3D_CHECKERBOARD:
189 fpa_type = 0;
190 break;
191 case AV_STEREO3D_COLUMNS:
192 fpa_type = 1;
193 break;
194 case AV_STEREO3D_LINES:
195 fpa_type = 2;
196 break;
197 case AV_STEREO3D_SIDEBYSIDE:
198 fpa_type = 3;
199 break;
200 case AV_STEREO3D_TOPBOTTOM:
201 fpa_type = 4;
202 break;
203 case AV_STEREO3D_FRAMESEQUENCE:
204 fpa_type = 5;
205 break;
206 default:
207 fpa_type = -1;
208 break;
209 }
210
211 if (fpa_type != x4->params.i_frame_packing) {
212 x4->params.i_frame_packing = fpa_type;
213 x264_encoder_reconfig(x4->enc, &x4->params);
214 }
215 }
216 }
217
218 static int X264_frame(AVCodecContext *ctx, AVPacket *pkt, const AVFrame *frame,
219 int *got_packet)
220 {
221 X264Context *x4 = ctx->priv_data;
222 x264_nal_t *nal;
223 int nnal, i, ret;
224 x264_picture_t pic_out;
225
226 x264_picture_init( &x4->pic );
227 x4->pic.img.i_csp = x4->params.i_csp;
228 if (x264_bit_depth > 8)
229 x4->pic.img.i_csp |= X264_CSP_HIGH_DEPTH;
230 x4->pic.img.i_plane = 3;
231
232 if (frame) {
233 for (i = 0; i < 3; i++) {
234 x4->pic.img.plane[i] = frame->data[i];
235 x4->pic.img.i_stride[i] = frame->linesize[i];
236 }
237
238 x4->pic.i_pts = frame->pts;
239
240 switch (frame->pict_type) {
241 case AV_PICTURE_TYPE_I:
242 x4->pic.i_type = x4->forced_idr ? X264_TYPE_IDR
243 : X264_TYPE_KEYFRAME;
244 break;
245 case AV_PICTURE_TYPE_P:
246 x4->pic.i_type = X264_TYPE_P;
247 break;
248 case AV_PICTURE_TYPE_B:
249 x4->pic.i_type = X264_TYPE_B;
250 break;
251 default:
252 x4->pic.i_type = X264_TYPE_AUTO;
253 break;
254 }
255 reconfig_encoder(ctx, frame);
256 }
257 do {
258 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
259 return AVERROR_UNKNOWN;
260
261 ret = encode_nals(ctx, pkt, nal, nnal);
262 if (ret < 0)
263 return ret;
264 } while (!ret && !frame && x264_encoder_delayed_frames(x4->enc));
265
266 pkt->pts = pic_out.i_pts;
267 pkt->dts = pic_out.i_dts;
268
269 #if FF_API_CODED_FRAME
270 FF_DISABLE_DEPRECATION_WARNINGS
271 switch (pic_out.i_type) {
272 case X264_TYPE_IDR:
273 case X264_TYPE_I:
274 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_I;
275 break;
276 case X264_TYPE_P:
277 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_P;
278 break;
279 case X264_TYPE_B:
280 case X264_TYPE_BREF:
281 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_B;
282 break;
283 }
284 FF_ENABLE_DEPRECATION_WARNINGS
285 #endif
286
287 pkt->flags |= AV_PKT_FLAG_KEY*pic_out.b_keyframe;
288 if (ret) {
289 uint8_t *sd = av_packet_new_side_data(pkt, AV_PKT_DATA_QUALITY_FACTOR,
290 sizeof(int));
291 if (!sd)
292 return AVERROR(ENOMEM);
293 *(int *)sd = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
294
295 #if FF_API_CODED_FRAME
296 FF_DISABLE_DEPRECATION_WARNINGS
297 ctx->coded_frame->quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
298 FF_ENABLE_DEPRECATION_WARNINGS
299 #endif
300 }
301
302 *got_packet = ret;
303 return 0;
304 }
305
306 static av_cold int X264_close(AVCodecContext *avctx)
307 {
308 X264Context *x4 = avctx->priv_data;
309
310 av_freep(&avctx->extradata);
311 av_freep(&x4->sei);
312
313 if (x4->enc) {
314 x264_encoder_close(x4->enc);
315 x4->enc = NULL;
316 }
317
318 return 0;
319 }
320
321 static int convert_pix_fmt(enum AVPixelFormat pix_fmt)
322 {
323 switch (pix_fmt) {
324 case AV_PIX_FMT_YUV420P:
325 case AV_PIX_FMT_YUVJ420P:
326 case AV_PIX_FMT_YUV420P9:
327 case AV_PIX_FMT_YUV420P10: return X264_CSP_I420;
328 case AV_PIX_FMT_YUV422P:
329 case AV_PIX_FMT_YUVJ422P:
330 case AV_PIX_FMT_YUV422P10: return X264_CSP_I422;
331 case AV_PIX_FMT_YUV444P:
332 case AV_PIX_FMT_YUVJ444P:
333 case AV_PIX_FMT_YUV444P9:
334 case AV_PIX_FMT_YUV444P10: return X264_CSP_I444;
335 case AV_PIX_FMT_NV12: return X264_CSP_NV12;
336 case AV_PIX_FMT_NV16:
337 case AV_PIX_FMT_NV20: return X264_CSP_NV16;
338 #ifdef X264_CSP_NV21
339 case AV_PIX_FMT_NV21: return X264_CSP_NV21;
340 #endif
341 };
342 return 0;
343 }
344
345 #define PARSE_X264_OPT(name, var)\
346 if (x4->var && x264_param_parse(&x4->params, name, x4->var) < 0) {\
347 av_log(avctx, AV_LOG_ERROR, "Error parsing option '%s' with value '%s'.\n", name, x4->var);\
348 return AVERROR(EINVAL);\
349 }
350
351 static av_cold int X264_init(AVCodecContext *avctx)
352 {
353 X264Context *x4 = avctx->priv_data;
354
355 #if CONFIG_LIBX262_ENCODER
356 if (avctx->codec_id == AV_CODEC_ID_MPEG2VIDEO) {
357 x4->params.b_mpeg2 = 1;
358 x264_param_default_mpeg2(&x4->params);
359 } else
360 #else
361 x264_param_default(&x4->params);
362 #endif
363
364 x4->params.b_deblocking_filter = avctx->flags & AV_CODEC_FLAG_LOOP_FILTER;
365
366 if (x4->preset || x4->tune)
367 if (x264_param_default_preset(&x4->params, x4->preset, x4->tune) < 0) {
368 av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", x4->preset, x4->tune);
369 return AVERROR(EINVAL);
370 }
371
372 if (avctx->level > 0)
373 x4->params.i_level_idc = avctx->level;
374
375 x4->params.pf_log = X264_log;
376 x4->params.p_log_private = avctx;
377 x4->params.i_log_level = X264_LOG_DEBUG;
378 x4->params.i_csp = convert_pix_fmt(avctx->pix_fmt);
379
380 if (avctx->bit_rate) {
381 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
382 x4->params.rc.i_rc_method = X264_RC_ABR;
383 }
384 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
385 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
386 x4->params.rc.b_stat_write = avctx->flags & AV_CODEC_FLAG_PASS1;
387 if (avctx->flags & AV_CODEC_FLAG_PASS2) {
388 x4->params.rc.b_stat_read = 1;
389 } else {
390 if (x4->crf >= 0) {
391 x4->params.rc.i_rc_method = X264_RC_CRF;
392 x4->params.rc.f_rf_constant = x4->crf;
393 } else if (x4->cqp >= 0) {
394 x4->params.rc.i_rc_method = X264_RC_CQP;
395 x4->params.rc.i_qp_constant = x4->cqp;
396 }
397
398 if (x4->crf_max >= 0)
399 x4->params.rc.f_rf_constant_max = x4->crf_max;
400 }
401
402 if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy > 0 &&
403 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
404 x4->params.rc.f_vbv_buffer_init =
405 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
406 }
407
408 if (avctx->i_quant_factor > 0)
409 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
410 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
411 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
412
413 if (avctx->gop_size >= 0)
414 x4->params.i_keyint_max = avctx->gop_size;
415 if (avctx->max_b_frames >= 0)
416 x4->params.i_bframe = avctx->max_b_frames;
417 if (avctx->scenechange_threshold >= 0)
418 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
419 if (avctx->qmin >= 0)
420 x4->params.rc.i_qp_min = avctx->qmin;
421 if (avctx->qmax >= 0)
422 x4->params.rc.i_qp_max = avctx->qmax;
423 if (avctx->max_qdiff >= 0)
424 x4->params.rc.i_qp_step = avctx->max_qdiff;
425 if (avctx->qblur >= 0)
426 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
427 if (avctx->qcompress >= 0)
428 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
429 if (avctx->refs >= 0)
430 x4->params.i_frame_reference = avctx->refs;
431 if (avctx->trellis >= 0)
432 x4->params.analyse.i_trellis = avctx->trellis;
433 if (avctx->me_range >= 0)
434 x4->params.analyse.i_me_range = avctx->me_range;
435 if (avctx->noise_reduction >= 0)
436 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
437 if (avctx->me_subpel_quality >= 0)
438 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
439 if (avctx->b_frame_strategy >= 0)
440 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
441 if (avctx->keyint_min >= 0)
442 x4->params.i_keyint_min = avctx->keyint_min;
443 if (avctx->coder_type >= 0)
444 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
445 if (avctx->me_cmp >= 0)
446 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
447
448 if (x4->aq_mode >= 0)
449 x4->params.rc.i_aq_mode = x4->aq_mode;
450 if (x4->aq_strength >= 0)
451 x4->params.rc.f_aq_strength = x4->aq_strength;
452 PARSE_X264_OPT("psy-rd", psy_rd);
453 PARSE_X264_OPT("deblock", deblock);
454 PARSE_X264_OPT("partitions", partitions);
455 PARSE_X264_OPT("stats", stats);
456 if (x4->psy >= 0)
457 x4->params.analyse.b_psy = x4->psy;
458 if (x4->rc_lookahead >= 0)
459 x4->params.rc.i_lookahead = x4->rc_lookahead;
460 if (x4->weightp >= 0)
461 x4->params.analyse.i_weighted_pred = x4->weightp;
462 if (x4->weightb >= 0)
463 x4->params.analyse.b_weighted_bipred = x4->weightb;
464 if (x4->cplxblur >= 0)
465 x4->params.rc.f_complexity_blur = x4->cplxblur;
466
467 if (x4->ssim >= 0)
468 x4->params.analyse.b_ssim = x4->ssim;
469 if (x4->intra_refresh >= 0)
470 x4->params.b_intra_refresh = x4->intra_refresh;
471 if (x4->bluray_compat >= 0) {
472 x4->params.b_bluray_compat = x4->bluray_compat;
473 x4->params.b_vfr_input = 0;
474 }
475 if (x4->b_bias != INT_MIN)
476 x4->params.i_bframe_bias = x4->b_bias;
477 if (x4->b_pyramid >= 0)
478 x4->params.i_bframe_pyramid = x4->b_pyramid;
479 if (x4->mixed_refs >= 0)
480 x4->params.analyse.b_mixed_references = x4->mixed_refs;
481 if (x4->dct8x8 >= 0)
482 x4->params.analyse.b_transform_8x8 = x4->dct8x8;
483 if (x4->fast_pskip >= 0)
484 x4->params.analyse.b_fast_pskip = x4->fast_pskip;
485 if (x4->aud >= 0)
486 x4->params.b_aud = x4->aud;
487 if (x4->mbtree >= 0)
488 x4->params.rc.b_mb_tree = x4->mbtree;
489 if (x4->direct_pred >= 0)
490 x4->params.analyse.i_direct_mv_pred = x4->direct_pred;
491
492 if (x4->slice_max_size >= 0)
493 x4->params.i_slice_max_size = x4->slice_max_size;
494
495 if (x4->fastfirstpass)
496 x264_param_apply_fastfirstpass(&x4->params);
497
498 if (x4->nal_hrd >= 0)
499 x4->params.i_nal_hrd = x4->nal_hrd;
500
501 if (x4->motion_est >= 0) {
502 x4->params.analyse.i_me_method = x4->motion_est;
503 #if FF_API_MOTION_EST
504 FF_DISABLE_DEPRECATION_WARNINGS
505 } else {
506 if (avctx->me_method == ME_EPZS)
507 x4->params.analyse.i_me_method = X264_ME_DIA;
508 else if (avctx->me_method == ME_HEX)
509 x4->params.analyse.i_me_method = X264_ME_HEX;
510 else if (avctx->me_method == ME_UMH)
511 x4->params.analyse.i_me_method = X264_ME_UMH;
512 else if (avctx->me_method == ME_FULL)
513 x4->params.analyse.i_me_method = X264_ME_ESA;
514 else if (avctx->me_method == ME_TESA)
515 x4->params.analyse.i_me_method = X264_ME_TESA;
516 FF_ENABLE_DEPRECATION_WARNINGS
517 #endif
518 }
519
520 if (x4->profile)
521 if (x264_param_apply_profile(&x4->params, x4->profile) < 0) {
522 av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile);
523 return AVERROR(EINVAL);
524 }
525
526 x4->params.i_width = avctx->width;
527 x4->params.i_height = avctx->height;
528 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
529 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
530 x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
531 x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
532
533 x4->params.analyse.b_psnr = avctx->flags & AV_CODEC_FLAG_PSNR;
534
535 x4->params.i_threads = avctx->thread_count;
536 if (avctx->thread_type)
537 x4->params.b_sliced_threads = avctx->thread_type == FF_THREAD_SLICE;
538
539 x4->params.b_interlaced = avctx->flags & AV_CODEC_FLAG_INTERLACED_DCT;
540
541 x4->params.b_open_gop = !(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
542
543 x4->params.i_slice_count = avctx->slices;
544
545 x4->params.vui.b_fullrange = avctx->pix_fmt == AV_PIX_FMT_YUVJ420P ||
546 avctx->pix_fmt == AV_PIX_FMT_YUVJ422P ||
547 avctx->pix_fmt == AV_PIX_FMT_YUVJ444P ||
548 avctx->color_range == AVCOL_RANGE_JPEG;
549
550 // x264 validates the values internally
551 x4->params.vui.i_colorprim = avctx->color_primaries;
552 x4->params.vui.i_transfer = avctx->color_trc;
553 x4->params.vui.i_colmatrix = avctx->colorspace;
554
555 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
556 x4->params.b_repeat_headers = 0;
557
558 if (x4->x264_params) {
559 AVDictionary *dict = NULL;
560 AVDictionaryEntry *en = NULL;
561
562 if (!av_dict_parse_string(&dict, x4->x264_params, "=", ":", 0)) {
563 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
564 if (x264_param_parse(&x4->params, en->key, en->value) < 0)
565 av_log(avctx, AV_LOG_WARNING,
566 "Error parsing option '%s = %s'.\n",
567 en->key, en->value);
568 }
569
570 av_dict_free(&dict);
571 }
572 }
573
574 // update AVCodecContext with x264 parameters
575 avctx->has_b_frames = x4->params.i_bframe ?
576 x4->params.i_bframe_pyramid ? 2 : 1 : 0;
577 if (avctx->max_b_frames < 0)
578 avctx->max_b_frames = 0;
579
580 avctx->bit_rate = x4->params.rc.i_bitrate*1000;
581
582 x4->enc = x264_encoder_open(&x4->params);
583 if (!x4->enc)
584 return AVERROR_UNKNOWN;
585
586 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
587 x264_nal_t *nal;
588 uint8_t *p;
589 int nnal, s, i;
590
591 s = x264_encoder_headers(x4->enc, &nal, &nnal);
592 avctx->extradata = p = av_mallocz(s + AV_INPUT_BUFFER_PADDING_SIZE);
593 if (!p)
594 return AVERROR(ENOMEM);
595
596 for (i = 0; i < nnal; i++) {
597 /* Don't put the SEI in extradata. */
598 if (nal[i].i_type == NAL_SEI) {
599 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
600 x4->sei_size = nal[i].i_payload;
601 x4->sei = av_malloc(x4->sei_size);
602 if (!x4->sei)
603 return AVERROR(ENOMEM);
604 memcpy(x4->sei, nal[i].p_payload, nal[i].i_payload);
605 continue;
606 }
607 memcpy(p, nal[i].p_payload, nal[i].i_payload);
608 p += nal[i].i_payload;
609 }
610 avctx->extradata_size = p - avctx->extradata;
611 }
612
613 return 0;
614 }
615
616 static const enum AVPixelFormat pix_fmts_8bit[] = {
617 AV_PIX_FMT_YUV420P,
618 AV_PIX_FMT_YUVJ420P,
619 AV_PIX_FMT_YUV422P,
620 AV_PIX_FMT_YUVJ422P,
621 AV_PIX_FMT_YUV444P,
622 AV_PIX_FMT_YUVJ444P,
623 AV_PIX_FMT_NV12,
624 AV_PIX_FMT_NV16,
625 #ifdef X264_CSP_NV21
626 AV_PIX_FMT_NV21,
627 #endif
628 AV_PIX_FMT_NONE
629 };
630 static const enum AVPixelFormat pix_fmts_9bit[] = {
631 AV_PIX_FMT_YUV420P9,
632 AV_PIX_FMT_YUV444P9,
633 AV_PIX_FMT_NONE
634 };
635 static const enum AVPixelFormat pix_fmts_10bit[] = {
636 AV_PIX_FMT_YUV420P10,
637 AV_PIX_FMT_YUV422P10,
638 AV_PIX_FMT_YUV444P10,
639 AV_PIX_FMT_NV20,
640 AV_PIX_FMT_NONE
641 };
642
643 static av_cold void X264_init_static(AVCodec *codec)
644 {
645 if (x264_bit_depth == 8)
646 codec->pix_fmts = pix_fmts_8bit;
647 else if (x264_bit_depth == 9)
648 codec->pix_fmts = pix_fmts_9bit;
649 else if (x264_bit_depth == 10)
650 codec->pix_fmts = pix_fmts_10bit;
651 }
652
653 #define OFFSET(x) offsetof(X264Context, x)
654 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
655 static const AVOption options[] = {
656 { "preset", "Set the encoding preset (cf. x264 --fullhelp)", OFFSET(preset), AV_OPT_TYPE_STRING, { .str = "medium" }, 0, 0, VE},
657 { "tune", "Tune the encoding params (cf. x264 --fullhelp)", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
658 { "profile", "Set profile restrictions (cf. x264 --fullhelp) ", OFFSET(profile), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
659 { "fastfirstpass", "Use fast settings when encoding first pass", OFFSET(fastfirstpass), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE},
660 { "crf", "Select the quality for constant quality mode", OFFSET(crf), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE },
661 { "crf_max", "In CRF mode, prevents VBV from lowering quality beyond this point.",OFFSET(crf_max), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE },
662 { "qp", "Constant quantization parameter rate control method",OFFSET(cqp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE },
663 { "aq-mode", "AQ method", OFFSET(aq_mode), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "aq_mode"},
664 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" },
665 { "variance", "Variance AQ (complexity mask)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_VARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
666 { "autovariance", "Auto-variance AQ (experimental)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOVARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
667 { "aq-strength", "AQ strength. Reduces blocking and blurring in flat and textured areas.", OFFSET(aq_strength), AV_OPT_TYPE_FLOAT, {.dbl = -1}, -1, FLT_MAX, VE},
668 { "psy", "Use psychovisual optimizations.", OFFSET(psy), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
669 { "psy-rd", "Strength of psychovisual optimization, in <psy-rd>:<psy-trellis> format.", OFFSET(psy_rd), AV_OPT_TYPE_STRING, {0 }, 0, 0, VE},
670 { "rc-lookahead", "Number of frames to look ahead for frametype and ratecontrol", OFFSET(rc_lookahead), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE },
671 { "weightb", "Weighted prediction for B-frames.", OFFSET(weightb), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
672 { "weightp", "Weighted prediction analysis method.", OFFSET(weightp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "weightp" },
673 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_NONE}, INT_MIN, INT_MAX, VE, "weightp" },
674 { "simple", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SIMPLE}, INT_MIN, INT_MAX, VE, "weightp" },
675 { "smart", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SMART}, INT_MIN, INT_MAX, VE, "weightp" },
676 { "ssim", "Calculate and print SSIM stats.", OFFSET(ssim), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
677 { "intra-refresh", "Use Periodic Intra Refresh instead of IDR frames.",OFFSET(intra_refresh),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
678 { "bluray-compat", "Bluray compatibility workarounds.", OFFSET(bluray_compat) ,AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
679 { "b-bias", "Influences how often B-frames are used", OFFSET(b_bias), AV_OPT_TYPE_INT, { .i64 = INT_MIN}, INT_MIN, INT_MAX, VE },
680 { "b-pyramid", "Keep some B-frames as references.", OFFSET(b_pyramid), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "b_pyramid" },
681 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NONE}, INT_MIN, INT_MAX, VE, "b_pyramid" },
682 { "strict", "Strictly hierarchical pyramid", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_STRICT}, INT_MIN, INT_MAX, VE, "b_pyramid" },
683 { "normal", "Non-strict (not Blu-ray compatible)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NORMAL}, INT_MIN, INT_MAX, VE, "b_pyramid" },
684 { "mixed-refs", "One reference per partition, as opposed to one reference per macroblock", OFFSET(mixed_refs), AV_OPT_TYPE_INT, { .i64 = -1}, -1, 1, VE },
685 { "8x8dct", "High profile 8x8 transform.", OFFSET(dct8x8), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
686 { "fast-pskip", NULL, OFFSET(fast_pskip), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
687 { "aud", "Use access unit delimiters.", OFFSET(aud), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
688 { "mbtree", "Use macroblock tree ratecontrol.", OFFSET(mbtree), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
689 { "deblock", "Loop filter parameters, in <alpha:beta> form.", OFFSET(deblock), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
690 { "cplxblur", "Reduce fluctuations in QP (before curve compression)", OFFSET(cplxblur), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE},
691 { "partitions", "A comma-separated list of partitions to consider. "
692 "Possible values: p8x8, p4x4, b8x8, i8x8, i4x4, none, all", OFFSET(partitions), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
693 { "direct-pred", "Direct MV prediction mode", OFFSET(direct_pred), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "direct-pred" },
694 { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_NONE }, 0, 0, VE, "direct-pred" },
695 { "spatial", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_SPATIAL }, 0, 0, VE, "direct-pred" },
696 { "temporal", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_TEMPORAL }, 0, 0, VE, "direct-pred" },
697 { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_AUTO }, 0, 0, VE, "direct-pred" },
698 { "slice-max-size","Limit the size of each slice in bytes", OFFSET(slice_max_size),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE },
699 { "stats", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
700 { "nal-hrd", "Signal HRD information (requires vbv-bufsize; "
701 "cbr not allowed in .mp4)", OFFSET(nal_hrd), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "nal-hrd" },
702 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_NONE}, INT_MIN, INT_MAX, VE, "nal-hrd" },
703 { "vbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_VBR}, INT_MIN, INT_MAX, VE, "nal-hrd" },
704 { "cbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_CBR}, INT_MIN, INT_MAX, VE, "nal-hrd" },
705 { "motion-est", "Set motion estimation method", OFFSET(motion_est), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, X264_ME_TESA, VE, "motion-est"},
706 { "dia", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_DIA }, INT_MIN, INT_MAX, VE, "motion-est" },
707 { "hex", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_HEX }, INT_MIN, INT_MAX, VE, "motion-est" },
708 { "umh", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_UMH }, INT_MIN, INT_MAX, VE, "motion-est" },
709 { "esa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_ESA }, INT_MIN, INT_MAX, VE, "motion-est" },
710 { "tesa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_TESA }, INT_MIN, INT_MAX, VE, "motion-est" },
711 { "forced-idr", "If forwarding iframes, require them to be IDR frames.", OFFSET(forced_idr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
712 { "x264-params", "Override the x264 configuration using a :-separated list of key=value parameters", OFFSET(x264_params), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
713 { NULL },
714 };
715
716 static const AVCodecDefault x264_defaults[] = {
717 { "b", "0" },
718 { "bf", "-1" },
719 { "g", "-1" },
720 { "i_qfactor", "-1" },
721 { "qmin", "-1" },
722 { "qmax", "-1" },
723 { "qdiff", "-1" },
724 { "qblur", "-1" },
725 { "qcomp", "-1" },
726 { "refs", "-1" },
727 { "sc_threshold", "-1" },
728 { "trellis", "-1" },
729 { "nr", "-1" },
730 { "me_range", "-1" },
731 #if FF_API_MOTION_EST
732 { "me_method", "-1" },
733 #endif
734 { "subq", "-1" },
735 { "b_strategy", "-1" },
736 { "keyint_min", "-1" },
737 { "coder", "-1" },
738 { "cmp", "-1" },
739 { "threads", AV_STRINGIFY(X264_THREADS_AUTO) },
740 { "thread_type", "0" },
741 { "flags", "+cgop" },
742 { "rc_init_occupancy","-1" },
743 { NULL },
744 };
745
746 #if CONFIG_LIBX264_ENCODER
747 static const AVClass class = {
748 .class_name = "libx264",
749 .item_name = av_default_item_name,
750 .option = options,
751 .version = LIBAVUTIL_VERSION_INT,
752 };
753
754 AVCodec ff_libx264_encoder = {
755 .name = "libx264",
756 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
757 .type = AVMEDIA_TYPE_VIDEO,
758 .id = AV_CODEC_ID_H264,
759 .priv_data_size = sizeof(X264Context),
760 .init = X264_init,
761 .encode2 = X264_frame,
762 .close = X264_close,
763 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
764 .priv_class = &class,
765 .defaults = x264_defaults,
766 .init_static_data = X264_init_static,
767 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
768 FF_CODEC_CAP_INIT_CLEANUP,
769 };
770 #endif
771
772 #if CONFIG_LIBX262_ENCODER
773 static const AVClass X262_class = {
774 .class_name = "libx262",
775 .item_name = av_default_item_name,
776 .option = options,
777 .version = LIBAVUTIL_VERSION_INT,
778 };
779
780 AVCodec ff_libx262_encoder = {
781 .name = "libx262",
782 .long_name = NULL_IF_CONFIG_SMALL("libx262 MPEG2VIDEO"),
783 .type = AVMEDIA_TYPE_VIDEO,
784 .id = AV_CODEC_ID_MPEG2VIDEO,
785 .priv_data_size = sizeof(X264Context),
786 .init = X264_init,
787 .encode2 = X264_frame,
788 .close = X264_close,
789 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
790 .priv_class = &X262_class,
791 .defaults = x264_defaults,
792 .pix_fmts = pix_fmts_8bit,
793 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
794 FF_CODEC_CAP_INIT_CLEANUP,
795 };
796 #endif