2 * H.264 encoding using the x264 library
3 * Copyright (C) 2005 Mans Rullgard <mans@mansr.com>
5 * This file is part of FFmpeg.
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.
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.
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
22 #include "libavutil/eval.h"
23 #include "libavutil/internal.h"
24 #include "libavutil/opt.h"
25 #include "libavutil/mem.h"
26 #include "libavutil/pixdesc.h"
27 #include "libavutil/stereo3d.h"
32 #define X264_API_IMPORTS 1
42 typedef struct X264Context {
88 static void X264_log(void *p, int level, const char *fmt, va_list args)
90 static const int level_map[] = {
91 [X264_LOG_ERROR] = AV_LOG_ERROR,
92 [X264_LOG_WARNING] = AV_LOG_WARNING,
93 [X264_LOG_INFO] = AV_LOG_INFO,
94 [X264_LOG_DEBUG] = AV_LOG_DEBUG
97 if (level < 0 || level > X264_LOG_DEBUG)
100 av_vlog(p, level_map[level], fmt, args);
104 static int encode_nals(AVCodecContext *ctx, AVPacket *pkt,
105 const x264_nal_t *nals, int nnal)
107 X264Context *x4 = ctx->priv_data;
109 int i, size = x4->sei_size, ret;
114 for (i = 0; i < nnal; i++)
115 size += nals[i].i_payload;
117 if ((ret = ff_alloc_packet2(ctx, pkt, size, 0)) < 0)
122 /* Write the SEI as part of the first frame. */
123 if (x4->sei_size > 0 && nnal > 0) {
124 if (x4->sei_size > size) {
125 av_log(ctx, AV_LOG_ERROR, "Error: nal buffer is too small\n");
128 memcpy(p, x4->sei, x4->sei_size);
134 for (i = 0; i < nnal; i++){
135 memcpy(p, nals[i].p_payload, nals[i].i_payload);
136 p += nals[i].i_payload;
142 static int avfmt2_num_planes(int avfmt)
145 case AV_PIX_FMT_YUV420P:
146 case AV_PIX_FMT_YUVJ420P:
147 case AV_PIX_FMT_YUV420P9:
148 case AV_PIX_FMT_YUV420P10:
149 case AV_PIX_FMT_YUV444P:
152 case AV_PIX_FMT_BGR0:
153 case AV_PIX_FMT_BGR24:
154 case AV_PIX_FMT_RGB24:
162 static void reconfig_encoder(AVCodecContext *ctx, const AVFrame *frame)
164 X264Context *x4 = ctx->priv_data;
165 AVFrameSideData *side_data;
168 if (x4->avcintra_class < 0) {
169 if (x4->params.b_interlaced && x4->params.b_tff != frame->top_field_first) {
171 x4->params.b_tff = frame->top_field_first;
172 x264_encoder_reconfig(x4->enc, &x4->params);
174 if (x4->params.vui.i_sar_height*ctx->sample_aspect_ratio.num != ctx->sample_aspect_ratio.den * x4->params.vui.i_sar_width) {
175 x4->params.vui.i_sar_height = ctx->sample_aspect_ratio.den;
176 x4->params.vui.i_sar_width = ctx->sample_aspect_ratio.num;
177 x264_encoder_reconfig(x4->enc, &x4->params);
180 if (x4->params.rc.i_vbv_buffer_size != ctx->rc_buffer_size / 1000 ||
181 x4->params.rc.i_vbv_max_bitrate != ctx->rc_max_rate / 1000) {
182 x4->params.rc.i_vbv_buffer_size = ctx->rc_buffer_size / 1000;
183 x4->params.rc.i_vbv_max_bitrate = ctx->rc_max_rate / 1000;
184 x264_encoder_reconfig(x4->enc, &x4->params);
187 if (x4->params.rc.i_rc_method == X264_RC_ABR &&
188 x4->params.rc.i_bitrate != ctx->bit_rate / 1000) {
189 x4->params.rc.i_bitrate = ctx->bit_rate / 1000;
190 x264_encoder_reconfig(x4->enc, &x4->params);
194 x4->params.rc.i_rc_method == X264_RC_CRF &&
195 x4->params.rc.f_rf_constant != x4->crf) {
196 x4->params.rc.f_rf_constant = x4->crf;
197 x264_encoder_reconfig(x4->enc, &x4->params);
200 if (x4->params.rc.i_rc_method == X264_RC_CQP &&
202 x4->params.rc.i_qp_constant != x4->cqp) {
203 x4->params.rc.i_qp_constant = x4->cqp;
204 x264_encoder_reconfig(x4->enc, &x4->params);
207 if (x4->crf_max >= 0 &&
208 x4->params.rc.f_rf_constant_max != x4->crf_max) {
209 x4->params.rc.f_rf_constant_max = x4->crf_max;
210 x264_encoder_reconfig(x4->enc, &x4->params);
214 side_data = av_frame_get_side_data(frame, AV_FRAME_DATA_STEREO3D);
216 AVStereo3D *stereo = (AVStereo3D *)side_data->data;
219 switch (stereo->type) {
220 case AV_STEREO3D_CHECKERBOARD:
223 case AV_STEREO3D_COLUMNS:
226 case AV_STEREO3D_LINES:
229 case AV_STEREO3D_SIDEBYSIDE:
232 case AV_STEREO3D_TOPBOTTOM:
235 case AV_STEREO3D_FRAMESEQUENCE:
243 if (fpa_type != x4->params.i_frame_packing) {
244 x4->params.i_frame_packing = fpa_type;
245 x264_encoder_reconfig(x4->enc, &x4->params);
250 static int X264_frame(AVCodecContext *ctx, AVPacket *pkt, const AVFrame *frame,
253 X264Context *x4 = ctx->priv_data;
256 x264_picture_t pic_out = {0};
259 x264_picture_init( &x4->pic );
260 x4->pic.img.i_csp = x4->params.i_csp;
261 if (x264_bit_depth > 8)
262 x4->pic.img.i_csp |= X264_CSP_HIGH_DEPTH;
263 x4->pic.img.i_plane = avfmt2_num_planes(ctx->pix_fmt);
266 for (i = 0; i < x4->pic.img.i_plane; i++) {
267 x4->pic.img.plane[i] = frame->data[i];
268 x4->pic.img.i_stride[i] = frame->linesize[i];
271 x4->pic.i_pts = frame->pts;
273 frame->pict_type == AV_PICTURE_TYPE_I ? X264_TYPE_KEYFRAME :
274 frame->pict_type == AV_PICTURE_TYPE_P ? X264_TYPE_P :
275 frame->pict_type == AV_PICTURE_TYPE_B ? X264_TYPE_B :
278 reconfig_encoder(ctx, frame);
281 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
282 return AVERROR_EXTERNAL;
284 ret = encode_nals(ctx, pkt, nal, nnal);
287 } while (!ret && !frame && x264_encoder_delayed_frames(x4->enc));
289 pkt->pts = pic_out.i_pts;
290 pkt->dts = pic_out.i_dts;
293 switch (pic_out.i_type) {
296 pict_type = AV_PICTURE_TYPE_I;
299 pict_type = AV_PICTURE_TYPE_P;
303 pict_type = AV_PICTURE_TYPE_B;
306 pict_type = AV_PICTURE_TYPE_NONE;
308 #if FF_API_CODED_FRAME
309 FF_DISABLE_DEPRECATION_WARNINGS
310 ctx->coded_frame->pict_type = pict_type;
311 FF_ENABLE_DEPRECATION_WARNINGS
314 pkt->flags |= AV_PKT_FLAG_KEY*pic_out.b_keyframe;
316 ff_side_data_set_encoder_stats(pkt, (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA, NULL, 0, pict_type);
318 #if FF_API_CODED_FRAME
319 FF_DISABLE_DEPRECATION_WARNINGS
320 ctx->coded_frame->quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
321 FF_ENABLE_DEPRECATION_WARNINGS
329 static av_cold int X264_close(AVCodecContext *avctx)
331 X264Context *x4 = avctx->priv_data;
333 av_freep(&avctx->extradata);
337 x264_encoder_close(x4->enc);
344 #define OPT_STR(opt, param) \
347 if (param && (ret = x264_param_parse(&x4->params, opt, param)) < 0) { \
348 if(ret == X264_PARAM_BAD_NAME) \
349 av_log(avctx, AV_LOG_ERROR, \
350 "bad option '%s': '%s'\n", opt, param); \
352 av_log(avctx, AV_LOG_ERROR, \
353 "bad value for '%s': '%s'\n", opt, param); \
358 static int convert_pix_fmt(enum AVPixelFormat pix_fmt)
361 case AV_PIX_FMT_YUV420P:
362 case AV_PIX_FMT_YUVJ420P:
363 case AV_PIX_FMT_YUV420P9:
364 case AV_PIX_FMT_YUV420P10: return X264_CSP_I420;
365 case AV_PIX_FMT_YUV422P:
366 case AV_PIX_FMT_YUVJ422P:
367 case AV_PIX_FMT_YUV422P10: return X264_CSP_I422;
368 case AV_PIX_FMT_YUV444P:
369 case AV_PIX_FMT_YUVJ444P:
370 case AV_PIX_FMT_YUV444P9:
371 case AV_PIX_FMT_YUV444P10: return X264_CSP_I444;
373 case AV_PIX_FMT_BGR0:
374 return X264_CSP_BGRA;
375 case AV_PIX_FMT_BGR24:
378 case AV_PIX_FMT_RGB24:
381 case AV_PIX_FMT_NV12: return X264_CSP_NV12;
382 case AV_PIX_FMT_NV16:
383 case AV_PIX_FMT_NV20: return X264_CSP_NV16;
388 #define PARSE_X264_OPT(name, var)\
389 if (x4->var && x264_param_parse(&x4->params, name, x4->var) < 0) {\
390 av_log(avctx, AV_LOG_ERROR, "Error parsing option '%s' with value '%s'.\n", name, x4->var);\
391 return AVERROR(EINVAL);\
394 static av_cold int X264_init(AVCodecContext *avctx)
396 X264Context *x4 = avctx->priv_data;
399 if (avctx->global_quality > 0)
400 av_log(avctx, AV_LOG_WARNING, "-qscale is ignored, -crf is recommended.\n");
402 #if CONFIG_LIBX262_ENCODER
403 if (avctx->codec_id == AV_CODEC_ID_MPEG2VIDEO) {
404 x4->params.b_mpeg2 = 1;
405 x264_param_default_mpeg2(&x4->params);
408 x264_param_default(&x4->params);
410 x4->params.b_deblocking_filter = avctx->flags & AV_CODEC_FLAG_LOOP_FILTER;
412 if (x4->preset || x4->tune)
413 if (x264_param_default_preset(&x4->params, x4->preset, x4->tune) < 0) {
415 av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", x4->preset, x4->tune);
416 av_log(avctx, AV_LOG_INFO, "Possible presets:");
417 for (i = 0; x264_preset_names[i]; i++)
418 av_log(avctx, AV_LOG_INFO, " %s", x264_preset_names[i]);
419 av_log(avctx, AV_LOG_INFO, "\n");
420 av_log(avctx, AV_LOG_INFO, "Possible tunes:");
421 for (i = 0; x264_tune_names[i]; i++)
422 av_log(avctx, AV_LOG_INFO, " %s", x264_tune_names[i]);
423 av_log(avctx, AV_LOG_INFO, "\n");
424 return AVERROR(EINVAL);
427 if (avctx->level > 0)
428 x4->params.i_level_idc = avctx->level;
430 x4->params.pf_log = X264_log;
431 x4->params.p_log_private = avctx;
432 x4->params.i_log_level = X264_LOG_DEBUG;
433 x4->params.i_csp = convert_pix_fmt(avctx->pix_fmt);
435 OPT_STR("weightp", x4->wpredp);
437 if (avctx->bit_rate) {
438 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
439 x4->params.rc.i_rc_method = X264_RC_ABR;
441 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
442 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
443 x4->params.rc.b_stat_write = avctx->flags & AV_CODEC_FLAG_PASS1;
444 if (avctx->flags & AV_CODEC_FLAG_PASS2) {
445 x4->params.rc.b_stat_read = 1;
448 x4->params.rc.i_rc_method = X264_RC_CRF;
449 x4->params.rc.f_rf_constant = x4->crf;
450 } else if (x4->cqp >= 0) {
451 x4->params.rc.i_rc_method = X264_RC_CQP;
452 x4->params.rc.i_qp_constant = x4->cqp;
455 if (x4->crf_max >= 0)
456 x4->params.rc.f_rf_constant_max = x4->crf_max;
459 if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy > 0 &&
460 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
461 x4->params.rc.f_vbv_buffer_init =
462 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
465 OPT_STR("level", x4->level);
467 if (avctx->i_quant_factor > 0)
468 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
469 if (avctx->b_quant_factor > 0)
470 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
471 if (avctx->chromaoffset)
472 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
474 if (avctx->gop_size >= 0)
475 x4->params.i_keyint_max = avctx->gop_size;
476 if (avctx->max_b_frames >= 0)
477 x4->params.i_bframe = avctx->max_b_frames;
478 if (avctx->scenechange_threshold >= 0)
479 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
480 if (avctx->qmin >= 0)
481 x4->params.rc.i_qp_min = avctx->qmin;
482 if (avctx->qmax >= 0)
483 x4->params.rc.i_qp_max = avctx->qmax;
484 if (avctx->max_qdiff >= 0)
485 x4->params.rc.i_qp_step = avctx->max_qdiff;
486 if (avctx->qblur >= 0)
487 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
488 if (avctx->qcompress >= 0)
489 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
490 if (avctx->refs >= 0)
491 x4->params.i_frame_reference = avctx->refs;
492 else if (x4->level) {
494 int mbn = FF_CEIL_RSHIFT(avctx->width, 4) * FF_CEIL_RSHIFT(avctx->height, 4);
497 int scale = X264_BUILD < 129 ? 384 : 1;
499 if (!strcmp(x4->level, "1b")) {
501 } else if (strlen(x4->level) <= 3){
502 level_id = av_strtod(x4->level, &tail) * 10 + 0.5;
507 av_log(avctx, AV_LOG_WARNING, "Failed to parse level\n");
509 for (i = 0; i<x264_levels[i].level_idc; i++)
510 if (x264_levels[i].level_idc == level_id)
511 x4->params.i_frame_reference = av_clip(x264_levels[i].dpb / mbn / scale, 1, x4->params.i_frame_reference);
514 if (avctx->trellis >= 0)
515 x4->params.analyse.i_trellis = avctx->trellis;
516 if (avctx->me_range >= 0)
517 x4->params.analyse.i_me_range = avctx->me_range;
518 if (avctx->noise_reduction >= 0)
519 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
520 if (avctx->me_subpel_quality >= 0)
521 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
522 if (avctx->b_frame_strategy >= 0)
523 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
524 if (avctx->keyint_min >= 0)
525 x4->params.i_keyint_min = avctx->keyint_min;
526 if (avctx->coder_type >= 0)
527 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
528 if (avctx->me_cmp >= 0)
529 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
531 if (x4->aq_mode >= 0)
532 x4->params.rc.i_aq_mode = x4->aq_mode;
533 if (x4->aq_strength >= 0)
534 x4->params.rc.f_aq_strength = x4->aq_strength;
535 PARSE_X264_OPT("psy-rd", psy_rd);
536 PARSE_X264_OPT("deblock", deblock);
537 PARSE_X264_OPT("partitions", partitions);
538 PARSE_X264_OPT("stats", stats);
540 x4->params.analyse.b_psy = x4->psy;
541 if (x4->rc_lookahead >= 0)
542 x4->params.rc.i_lookahead = x4->rc_lookahead;
543 if (x4->weightp >= 0)
544 x4->params.analyse.i_weighted_pred = x4->weightp;
545 if (x4->weightb >= 0)
546 x4->params.analyse.b_weighted_bipred = x4->weightb;
547 if (x4->cplxblur >= 0)
548 x4->params.rc.f_complexity_blur = x4->cplxblur;
551 x4->params.analyse.b_ssim = x4->ssim;
552 if (x4->intra_refresh >= 0)
553 x4->params.b_intra_refresh = x4->intra_refresh;
554 if (x4->bluray_compat >= 0) {
555 x4->params.b_bluray_compat = x4->bluray_compat;
556 x4->params.b_vfr_input = 0;
558 if (x4->avcintra_class >= 0)
559 #if X264_BUILD >= 142
560 x4->params.i_avcintra_class = x4->avcintra_class;
562 av_log(avctx, AV_LOG_ERROR,
563 "x264 too old for AVC Intra, at least version 142 needed\n");
565 if (x4->b_bias != INT_MIN)
566 x4->params.i_bframe_bias = x4->b_bias;
567 if (x4->b_pyramid >= 0)
568 x4->params.i_bframe_pyramid = x4->b_pyramid;
569 if (x4->mixed_refs >= 0)
570 x4->params.analyse.b_mixed_references = x4->mixed_refs;
572 x4->params.analyse.b_transform_8x8 = x4->dct8x8;
573 if (x4->fast_pskip >= 0)
574 x4->params.analyse.b_fast_pskip = x4->fast_pskip;
576 x4->params.b_aud = x4->aud;
578 x4->params.rc.b_mb_tree = x4->mbtree;
579 if (x4->direct_pred >= 0)
580 x4->params.analyse.i_direct_mv_pred = x4->direct_pred;
582 if (x4->slice_max_size >= 0)
583 x4->params.i_slice_max_size = x4->slice_max_size;
586 * Allow x264 to be instructed through AVCodecContext about the maximum
587 * size of the RTP payload. For example, this enables the production of
588 * payload suitable for the H.264 RTP packetization-mode 0 i.e. single
589 * NAL unit per RTP packet.
591 if (avctx->rtp_payload_size)
592 x4->params.i_slice_max_size = avctx->rtp_payload_size;
595 if (x4->fastfirstpass)
596 x264_param_apply_fastfirstpass(&x4->params);
598 /* Allow specifying the x264 profile through AVCodecContext. */
600 switch (avctx->profile) {
601 case FF_PROFILE_H264_BASELINE:
602 x4->profile = av_strdup("baseline");
604 case FF_PROFILE_H264_HIGH:
605 x4->profile = av_strdup("high");
607 case FF_PROFILE_H264_HIGH_10:
608 x4->profile = av_strdup("high10");
610 case FF_PROFILE_H264_HIGH_422:
611 x4->profile = av_strdup("high422");
613 case FF_PROFILE_H264_HIGH_444:
614 x4->profile = av_strdup("high444");
616 case FF_PROFILE_H264_MAIN:
617 x4->profile = av_strdup("main");
623 if (x4->nal_hrd >= 0)
624 x4->params.i_nal_hrd = x4->nal_hrd;
626 if (x4->motion_est >= 0) {
627 x4->params.analyse.i_me_method = x4->motion_est;
628 #if FF_API_MOTION_EST
629 FF_DISABLE_DEPRECATION_WARNINGS
631 if (avctx->me_method == ME_EPZS)
632 x4->params.analyse.i_me_method = X264_ME_DIA;
633 else if (avctx->me_method == ME_HEX)
634 x4->params.analyse.i_me_method = X264_ME_HEX;
635 else if (avctx->me_method == ME_UMH)
636 x4->params.analyse.i_me_method = X264_ME_UMH;
637 else if (avctx->me_method == ME_FULL)
638 x4->params.analyse.i_me_method = X264_ME_ESA;
639 else if (avctx->me_method == ME_TESA)
640 x4->params.analyse.i_me_method = X264_ME_TESA;
641 FF_ENABLE_DEPRECATION_WARNINGS
646 if (x264_param_apply_profile(&x4->params, x4->profile) < 0) {
648 av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile);
649 av_log(avctx, AV_LOG_INFO, "Possible profiles:");
650 for (i = 0; x264_profile_names[i]; i++)
651 av_log(avctx, AV_LOG_INFO, " %s", x264_profile_names[i]);
652 av_log(avctx, AV_LOG_INFO, "\n");
653 return AVERROR(EINVAL);
656 x4->params.i_width = avctx->width;
657 x4->params.i_height = avctx->height;
658 av_reduce(&sw, &sh, avctx->sample_aspect_ratio.num, avctx->sample_aspect_ratio.den, 4096);
659 x4->params.vui.i_sar_width = sw;
660 x4->params.vui.i_sar_height = sh;
661 x4->params.i_timebase_den = avctx->time_base.den;
662 x4->params.i_timebase_num = avctx->time_base.num;
663 x4->params.i_fps_num = avctx->time_base.den;
664 x4->params.i_fps_den = avctx->time_base.num * avctx->ticks_per_frame;
666 x4->params.analyse.b_psnr = avctx->flags & AV_CODEC_FLAG_PSNR;
668 x4->params.i_threads = avctx->thread_count;
669 if (avctx->thread_type)
670 x4->params.b_sliced_threads = avctx->thread_type == FF_THREAD_SLICE;
672 x4->params.b_interlaced = avctx->flags & AV_CODEC_FLAG_INTERLACED_DCT;
674 x4->params.b_open_gop = !(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
676 x4->params.i_slice_count = avctx->slices;
678 x4->params.vui.b_fullrange = avctx->pix_fmt == AV_PIX_FMT_YUVJ420P ||
679 avctx->pix_fmt == AV_PIX_FMT_YUVJ422P ||
680 avctx->pix_fmt == AV_PIX_FMT_YUVJ444P ||
681 avctx->color_range == AVCOL_RANGE_JPEG;
683 if (avctx->colorspace != AVCOL_SPC_UNSPECIFIED)
684 x4->params.vui.i_colmatrix = avctx->colorspace;
685 if (avctx->color_primaries != AVCOL_PRI_UNSPECIFIED)
686 x4->params.vui.i_colorprim = avctx->color_primaries;
687 if (avctx->color_trc != AVCOL_TRC_UNSPECIFIED)
688 x4->params.vui.i_transfer = avctx->color_trc;
690 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
691 x4->params.b_repeat_headers = 0;
694 const char *p= x4->x264opts;
696 char param[256]={0}, val[256]={0};
697 if(sscanf(p, "%255[^:=]=%255[^:]", param, val) == 1){
706 if (x4->x264_params) {
707 AVDictionary *dict = NULL;
708 AVDictionaryEntry *en = NULL;
710 if (!av_dict_parse_string(&dict, x4->x264_params, "=", ":", 0)) {
711 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
712 if (x264_param_parse(&x4->params, en->key, en->value) < 0)
713 av_log(avctx, AV_LOG_WARNING,
714 "Error parsing option '%s = %s'.\n",
722 // update AVCodecContext with x264 parameters
723 avctx->has_b_frames = x4->params.i_bframe ?
724 x4->params.i_bframe_pyramid ? 2 : 1 : 0;
725 if (avctx->max_b_frames < 0)
726 avctx->max_b_frames = 0;
728 avctx->bit_rate = x4->params.rc.i_bitrate*1000;
730 x4->enc = x264_encoder_open(&x4->params);
732 return AVERROR_EXTERNAL;
734 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
739 s = x264_encoder_headers(x4->enc, &nal, &nnal);
740 avctx->extradata = p = av_malloc(s);
742 return AVERROR(ENOMEM);
744 for (i = 0; i < nnal; i++) {
745 /* Don't put the SEI in extradata. */
746 if (nal[i].i_type == NAL_SEI) {
747 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
748 x4->sei_size = nal[i].i_payload;
749 x4->sei = av_malloc(x4->sei_size);
751 return AVERROR(ENOMEM);
752 memcpy(x4->sei, nal[i].p_payload, nal[i].i_payload);
755 memcpy(p, nal[i].p_payload, nal[i].i_payload);
756 p += nal[i].i_payload;
758 avctx->extradata_size = p - avctx->extradata;
764 static const enum AVPixelFormat pix_fmts_8bit[] = {
775 static const enum AVPixelFormat pix_fmts_9bit[] = {
780 static const enum AVPixelFormat pix_fmts_10bit[] = {
781 AV_PIX_FMT_YUV420P10,
782 AV_PIX_FMT_YUV422P10,
783 AV_PIX_FMT_YUV444P10,
787 static const enum AVPixelFormat pix_fmts_8bit_rgb[] = {
796 static av_cold void X264_init_static(AVCodec *codec)
798 if (x264_bit_depth == 8)
799 codec->pix_fmts = pix_fmts_8bit;
800 else if (x264_bit_depth == 9)
801 codec->pix_fmts = pix_fmts_9bit;
802 else if (x264_bit_depth == 10)
803 codec->pix_fmts = pix_fmts_10bit;
806 #define OFFSET(x) offsetof(X264Context, x)
807 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
808 static const AVOption options[] = {
809 { "preset", "Set the encoding preset (cf. x264 --fullhelp)", OFFSET(preset), AV_OPT_TYPE_STRING, { .str = "medium" }, 0, 0, VE},
810 { "tune", "Tune the encoding params (cf. x264 --fullhelp)", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
811 { "profile", "Set profile restrictions (cf. x264 --fullhelp) ", OFFSET(profile), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
812 { "fastfirstpass", "Use fast settings when encoding first pass", OFFSET(fastfirstpass), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE},
813 {"level", "Specify level (as defined by Annex A)", OFFSET(level), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
814 {"passlogfile", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
815 {"wpredp", "Weighted prediction for P-frames", OFFSET(wpredp), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
816 {"x264opts", "x264 options", OFFSET(x264opts), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
817 { "crf", "Select the quality for constant quality mode", OFFSET(crf), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE },
818 { "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 },
819 { "qp", "Constant quantization parameter rate control method",OFFSET(cqp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE },
820 { "aq-mode", "AQ method", OFFSET(aq_mode), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "aq_mode"},
821 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" },
822 { "variance", "Variance AQ (complexity mask)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_VARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
823 { "autovariance", "Auto-variance AQ", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOVARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
824 #if X264_BUILD >= 144
825 { "autovariance-biased", "Auto-variance AQ with bias to dark scenes", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOVARIANCE_BIASED}, INT_MIN, INT_MAX, VE, "aq_mode" },
827 { "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},
828 { "psy", "Use psychovisual optimizations.", OFFSET(psy), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
829 { "psy-rd", "Strength of psychovisual optimization, in <psy-rd>:<psy-trellis> format.", OFFSET(psy_rd), AV_OPT_TYPE_STRING, {0 }, 0, 0, VE},
830 { "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 },
831 { "weightb", "Weighted prediction for B-frames.", OFFSET(weightb), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
832 { "weightp", "Weighted prediction analysis method.", OFFSET(weightp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "weightp" },
833 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_NONE}, INT_MIN, INT_MAX, VE, "weightp" },
834 { "simple", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SIMPLE}, INT_MIN, INT_MAX, VE, "weightp" },
835 { "smart", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SMART}, INT_MIN, INT_MAX, VE, "weightp" },
836 { "ssim", "Calculate and print SSIM stats.", OFFSET(ssim), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
837 { "intra-refresh", "Use Periodic Intra Refresh instead of IDR frames.",OFFSET(intra_refresh),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
838 { "bluray-compat", "Bluray compatibility workarounds.", OFFSET(bluray_compat) ,AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
839 { "b-bias", "Influences how often B-frames are used", OFFSET(b_bias), AV_OPT_TYPE_INT, { .i64 = INT_MIN}, INT_MIN, INT_MAX, VE },
840 { "b-pyramid", "Keep some B-frames as references.", OFFSET(b_pyramid), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "b_pyramid" },
841 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NONE}, INT_MIN, INT_MAX, VE, "b_pyramid" },
842 { "strict", "Strictly hierarchical pyramid", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_STRICT}, INT_MIN, INT_MAX, VE, "b_pyramid" },
843 { "normal", "Non-strict (not Blu-ray compatible)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NORMAL}, INT_MIN, INT_MAX, VE, "b_pyramid" },
844 { "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 },
845 { "8x8dct", "High profile 8x8 transform.", OFFSET(dct8x8), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
846 { "fast-pskip", NULL, OFFSET(fast_pskip), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
847 { "aud", "Use access unit delimiters.", OFFSET(aud), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
848 { "mbtree", "Use macroblock tree ratecontrol.", OFFSET(mbtree), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE},
849 { "deblock", "Loop filter parameters, in <alpha:beta> form.", OFFSET(deblock), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
850 { "cplxblur", "Reduce fluctuations in QP (before curve compression)", OFFSET(cplxblur), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE},
851 { "partitions", "A comma-separated list of partitions to consider. "
852 "Possible values: p8x8, p4x4, b8x8, i8x8, i4x4, none, all", OFFSET(partitions), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
853 { "direct-pred", "Direct MV prediction mode", OFFSET(direct_pred), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "direct-pred" },
854 { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_NONE }, 0, 0, VE, "direct-pred" },
855 { "spatial", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_SPATIAL }, 0, 0, VE, "direct-pred" },
856 { "temporal", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_TEMPORAL }, 0, 0, VE, "direct-pred" },
857 { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_AUTO }, 0, 0, VE, "direct-pred" },
858 { "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 },
859 { "stats", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
860 { "nal-hrd", "Signal HRD information (requires vbv-bufsize; "
861 "cbr not allowed in .mp4)", OFFSET(nal_hrd), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "nal-hrd" },
862 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_NONE}, INT_MIN, INT_MAX, VE, "nal-hrd" },
863 { "vbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_VBR}, INT_MIN, INT_MAX, VE, "nal-hrd" },
864 { "cbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_CBR}, INT_MIN, INT_MAX, VE, "nal-hrd" },
865 { "avcintra-class","AVC-Intra class 50/100/200", OFFSET(avcintra_class),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 200 , VE},
866 { "motion-est", "Set motion estimation method", OFFSET(motion_est), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, X264_ME_TESA, VE, "motion-est"},
867 { "dia", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_DIA }, INT_MIN, INT_MAX, VE, "motion-est" },
868 { "hex", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_HEX }, INT_MIN, INT_MAX, VE, "motion-est" },
869 { "umh", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_UMH }, INT_MIN, INT_MAX, VE, "motion-est" },
870 { "esa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_ESA }, INT_MIN, INT_MAX, VE, "motion-est" },
871 { "tesa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_ME_TESA }, INT_MIN, INT_MAX, VE, "motion-est" },
872 { "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 },
876 static const AVCodecDefault x264_defaults[] = {
881 { "i_qfactor", "-1" },
882 { "b_qfactor", "-1" },
888 // { "rc_lookahead", "-1" },
890 { "sc_threshold", "-1" },
893 { "me_range", "-1" },
894 #if FF_API_MOTION_EST
895 { "me_method", "-1" },
898 { "b_strategy", "-1" },
899 { "keyint_min", "-1" },
902 { "threads", AV_STRINGIFY(X264_THREADS_AUTO) },
903 { "thread_type", "0" },
904 { "flags", "+cgop" },
905 { "rc_init_occupancy","-1" },
909 #if CONFIG_LIBX264_ENCODER
910 static const AVClass x264_class = {
911 .class_name = "libx264",
912 .item_name = av_default_item_name,
914 .version = LIBAVUTIL_VERSION_INT,
917 static const AVClass rgbclass = {
918 .class_name = "libx264rgb",
919 .item_name = av_default_item_name,
921 .version = LIBAVUTIL_VERSION_INT,
924 AVCodec ff_libx264_encoder = {
926 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
927 .type = AVMEDIA_TYPE_VIDEO,
928 .id = AV_CODEC_ID_H264,
929 .priv_data_size = sizeof(X264Context),
931 .encode2 = X264_frame,
933 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
934 .priv_class = &x264_class,
935 .defaults = x264_defaults,
936 .init_static_data = X264_init_static,
937 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
938 FF_CODEC_CAP_INIT_CLEANUP,
941 AVCodec ff_libx264rgb_encoder = {
942 .name = "libx264rgb",
943 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 RGB"),
944 .type = AVMEDIA_TYPE_VIDEO,
945 .id = AV_CODEC_ID_H264,
946 .priv_data_size = sizeof(X264Context),
948 .encode2 = X264_frame,
950 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
951 .priv_class = &rgbclass,
952 .defaults = x264_defaults,
953 .pix_fmts = pix_fmts_8bit_rgb,
957 #if CONFIG_LIBX262_ENCODER
958 static const AVClass X262_class = {
959 .class_name = "libx262",
960 .item_name = av_default_item_name,
962 .version = LIBAVUTIL_VERSION_INT,
965 AVCodec ff_libx262_encoder = {
967 .long_name = NULL_IF_CONFIG_SMALL("libx262 MPEG2VIDEO"),
968 .type = AVMEDIA_TYPE_VIDEO,
969 .id = AV_CODEC_ID_MPEG2VIDEO,
970 .priv_data_size = sizeof(X264Context),
972 .encode2 = X264_frame,
974 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
975 .priv_class = &X262_class,
976 .defaults = x264_defaults,
977 .pix_fmts = pix_fmts_8bit,
978 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE |
979 FF_CODEC_CAP_INIT_CLEANUP,