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 FFmpeg. 6 * 7 * FFmpeg is free software; you can redistribute it and/or 8 * modify it under the terms of the GNU Lesser General Public 9 * License as published by the Free Software Foundation; either 10 * version 2.1 of the License, or (at your option) any later version. 11 * 12 * FFmpeg is distributed in the hope that it will be useful, 13 * but WITHOUT ANY WARRANTY; without even the implied warranty of 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 15 * Lesser General Public License for more details. 16 * 17 * You should have received a copy of the GNU Lesser General Public 18 * License along with FFmpeg; if not, write to the Free Software 19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 20 */ 21 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" 28#include "avcodec.h" 29#include "internal.h" 30 31#if defined(_MSC_VER) 32#define X264_API_IMPORTS 1 33#endif 34 35#include <x264.h> 36#include <float.h> 37#include <math.h> 38#include <stdio.h> 39#include <stdlib.h> 40#include <string.h> 41 42typedef struct X264Context { 43 AVClass *class; 44 x264_param_t params; 45 x264_t *enc; 46 x264_picture_t pic; 47 uint8_t *sei; 48 int sei_size; 49 char *preset; 50 char *tune; 51 char *profile; 52 char *level; 53 int fastfirstpass; 54 char *wpredp; 55 char *x264opts; 56 float crf; 57 float crf_max; 58 int cqp; 59 int aq_mode; 60 float aq_strength; 61 char *psy_rd; 62 int psy; 63 int rc_lookahead; 64 int weightp; 65 int weightb; 66 int ssim; 67 int intra_refresh; 68 int bluray_compat; 69 int b_bias; 70 int b_pyramid; 71 int mixed_refs; 72 int dct8x8; 73 int fast_pskip; 74 int aud; 75 int mbtree; 76 char *deblock; 77 float cplxblur; 78 char *partitions; 79 int direct_pred; 80 int slice_max_size; 81 char *stats; 82 int nal_hrd; 83 char *x264_params; 84} X264Context; 85 86static void X264_log(void *p, int level, const char *fmt, va_list args) 87{ 88 static const int level_map[] = { 89 [X264_LOG_ERROR] = AV_LOG_ERROR, 90 [X264_LOG_WARNING] = AV_LOG_WARNING, 91 [X264_LOG_INFO] = AV_LOG_INFO, 92 [X264_LOG_DEBUG] = AV_LOG_DEBUG 93 }; 94 95 if (level < 0 || level > X264_LOG_DEBUG) 96 return; 97 98 av_vlog(p, level_map[level], fmt, args); 99} 100 101 102static int encode_nals(AVCodecContext *ctx, AVPacket *pkt, 103 x264_nal_t *nals, int nnal) 104{ 105 X264Context *x4 = ctx->priv_data; 106 uint8_t *p; 107 int i, size = x4->sei_size, ret; 108 109 if (!nnal) 110 return 0; 111 112 for (i = 0; i < nnal; i++) 113 size += nals[i].i_payload; 114 115 if ((ret = ff_alloc_packet2(ctx, pkt, size)) < 0) 116 return ret; 117 118 p = pkt->data; 119 120 /* Write the SEI as part of the first frame. */ 121 if (x4->sei_size > 0 && nnal > 0) { 122 if (x4->sei_size > size) { 123 av_log(ctx, AV_LOG_ERROR, "Error: nal buffer is too small\n"); 124 return -1; 125 } 126 memcpy(p, x4->sei, x4->sei_size); 127 p += x4->sei_size; 128 x4->sei_size = 0; 129 av_freep(&x4->sei); 130 } 131 132 for (i = 0; i < nnal; i++){ 133 memcpy(p, nals[i].p_payload, nals[i].i_payload); 134 p += nals[i].i_payload; 135 } 136 137 return 1; 138} 139 140static int avfmt2_num_planes(int avfmt) 141{ 142 switch (avfmt) { 143 case AV_PIX_FMT_YUV420P: 144 case AV_PIX_FMT_YUVJ420P: 145 case AV_PIX_FMT_YUV420P9: 146 case AV_PIX_FMT_YUV420P10: 147 case AV_PIX_FMT_YUV444P: 148 return 3; 149 150 case AV_PIX_FMT_BGR24: 151 case AV_PIX_FMT_RGB24: 152 return 1; 153 154 default: 155 return 3; 156 } 157} 158 159static int X264_frame(AVCodecContext *ctx, AVPacket *pkt, const AVFrame *frame, 160 int *got_packet) 161{ 162 X264Context *x4 = ctx->priv_data; 163 x264_nal_t *nal; 164 int nnal, i, ret; 165 x264_picture_t pic_out = {0}; 166 AVFrameSideData *side_data; 167 168 x264_picture_init( &x4->pic ); 169 x4->pic.img.i_csp = x4->params.i_csp; 170 if (x264_bit_depth > 8) 171 x4->pic.img.i_csp |= X264_CSP_HIGH_DEPTH; 172 x4->pic.img.i_plane = avfmt2_num_planes(ctx->pix_fmt); 173 174 if (frame) { 175 for (i = 0; i < x4->pic.img.i_plane; i++) { 176 x4->pic.img.plane[i] = frame->data[i]; 177 x4->pic.img.i_stride[i] = frame->linesize[i]; 178 } 179 180 x4->pic.i_pts = frame->pts; 181 x4->pic.i_type = 182 frame->pict_type == AV_PICTURE_TYPE_I ? X264_TYPE_KEYFRAME : 183 frame->pict_type == AV_PICTURE_TYPE_P ? X264_TYPE_P : 184 frame->pict_type == AV_PICTURE_TYPE_B ? X264_TYPE_B : 185 X264_TYPE_AUTO; 186 if (x4->params.b_interlaced && x4->params.b_tff != frame->top_field_first) { 187 x4->params.b_tff = frame->top_field_first; 188 x264_encoder_reconfig(x4->enc, &x4->params); 189 } 190 if (x4->params.vui.i_sar_height != ctx->sample_aspect_ratio.den || 191 x4->params.vui.i_sar_width != ctx->sample_aspect_ratio.num) { 192 x4->params.vui.i_sar_height = ctx->sample_aspect_ratio.den; 193 x4->params.vui.i_sar_width = ctx->sample_aspect_ratio.num; 194 x264_encoder_reconfig(x4->enc, &x4->params); 195 } 196 197 if (x4->params.rc.i_vbv_buffer_size != ctx->rc_buffer_size / 1000 || 198 x4->params.rc.i_vbv_max_bitrate != ctx->rc_max_rate / 1000) { 199 x4->params.rc.i_vbv_buffer_size = ctx->rc_buffer_size / 1000; 200 x4->params.rc.i_vbv_max_bitrate = ctx->rc_max_rate / 1000; 201 x264_encoder_reconfig(x4->enc, &x4->params); 202 } 203 204 if (x4->params.rc.i_rc_method == X264_RC_ABR && 205 x4->params.rc.i_bitrate != ctx->bit_rate / 1000) { 206 x4->params.rc.i_bitrate = ctx->bit_rate / 1000; 207 x264_encoder_reconfig(x4->enc, &x4->params); 208 } 209 210 if (x4->crf >= 0 && 211 x4->params.rc.i_rc_method == X264_RC_CRF && 212 x4->params.rc.f_rf_constant != x4->crf) { 213 x4->params.rc.f_rf_constant = x4->crf; 214 x264_encoder_reconfig(x4->enc, &x4->params); 215 } 216 217 if (x4->params.rc.i_rc_method == X264_RC_CQP && 218 x4->cqp >= 0 && 219 x4->params.rc.i_qp_constant != x4->cqp) { 220 x4->params.rc.i_qp_constant = x4->cqp; 221 x264_encoder_reconfig(x4->enc, &x4->params); 222 } 223 224 if (x4->crf_max >= 0 && 225 x4->params.rc.f_rf_constant_max != x4->crf_max) { 226 x4->params.rc.f_rf_constant_max = x4->crf_max; 227 x264_encoder_reconfig(x4->enc, &x4->params); 228 } 229 230 side_data = av_frame_get_side_data(frame, AV_FRAME_DATA_STEREO3D); 231 if (side_data) { 232 AVStereo3D *stereo = (AVStereo3D *)side_data->data; 233 int fpa_type; 234 235 switch (stereo->type) { 236 case AV_STEREO3D_CHECKERBOARD: 237 fpa_type = 0; 238 break; 239 case AV_STEREO3D_COLUMNS: 240 fpa_type = 1; 241 break; 242 case AV_STEREO3D_LINES: 243 fpa_type = 2; 244 break; 245 case AV_STEREO3D_SIDEBYSIDE: 246 fpa_type = 3; 247 break; 248 case AV_STEREO3D_TOPBOTTOM: 249 fpa_type = 4; 250 break; 251 case AV_STEREO3D_FRAMESEQUENCE: 252 fpa_type = 5; 253 break; 254 default: 255 fpa_type = -1; 256 break; 257 } 258 259 if (fpa_type != x4->params.i_frame_packing) { 260 x4->params.i_frame_packing = fpa_type; 261 x264_encoder_reconfig(x4->enc, &x4->params); 262 } 263 } 264 } 265 do { 266 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0) 267 return -1; 268 269 ret = encode_nals(ctx, pkt, nal, nnal); 270 if (ret < 0) 271 return -1; 272 } while (!ret && !frame && x264_encoder_delayed_frames(x4->enc)); 273 274 pkt->pts = pic_out.i_pts; 275 pkt->dts = pic_out.i_dts; 276 277 switch (pic_out.i_type) { 278 case X264_TYPE_IDR: 279 case X264_TYPE_I: 280 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_I; 281 break; 282 case X264_TYPE_P: 283 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_P; 284 break; 285 case X264_TYPE_B: 286 case X264_TYPE_BREF: 287 ctx->coded_frame->pict_type = AV_PICTURE_TYPE_B; 288 break; 289 } 290 291 pkt->flags |= AV_PKT_FLAG_KEY*pic_out.b_keyframe; 292 if (ret) 293 ctx->coded_frame->quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA; 294 295 *got_packet = ret; 296 return 0; 297} 298 299static av_cold int X264_close(AVCodecContext *avctx) 300{ 301 X264Context *x4 = avctx->priv_data; 302 303 av_freep(&avctx->extradata); 304 av_free(x4->sei); 305 306 if (x4->enc) 307 x264_encoder_close(x4->enc); 308 309 av_frame_free(&avctx->coded_frame); 310 311 return 0; 312} 313 314#define OPT_STR(opt, param) \ 315 do { \ 316 int ret; \ 317 if (param!=NULL && (ret = x264_param_parse(&x4->params, opt, param)) < 0) { \ 318 if(ret == X264_PARAM_BAD_NAME) \ 319 av_log(avctx, AV_LOG_ERROR, \ 320 "bad option '%s': '%s'\n", opt, param); \ 321 else \ 322 av_log(avctx, AV_LOG_ERROR, \ 323 "bad value for '%s': '%s'\n", opt, param); \ 324 return -1; \ 325 } \ 326 } while (0) 327 328static int convert_pix_fmt(enum AVPixelFormat pix_fmt) 329{ 330 switch (pix_fmt) { 331 case AV_PIX_FMT_YUV420P: 332 case AV_PIX_FMT_YUVJ420P: 333 case AV_PIX_FMT_YUV420P9: 334 case AV_PIX_FMT_YUV420P10: return X264_CSP_I420; 335 case AV_PIX_FMT_YUV422P: 336 case AV_PIX_FMT_YUVJ422P: 337 case AV_PIX_FMT_YUV422P10: return X264_CSP_I422; 338 case AV_PIX_FMT_YUV444P: 339 case AV_PIX_FMT_YUVJ444P: 340 case AV_PIX_FMT_YUV444P9: 341 case AV_PIX_FMT_YUV444P10: return X264_CSP_I444; 342#ifdef X264_CSP_BGR 343 case AV_PIX_FMT_BGR24: 344 return X264_CSP_BGR; 345 346 case AV_PIX_FMT_RGB24: 347 return X264_CSP_RGB; 348#endif 349 case AV_PIX_FMT_NV12: return X264_CSP_NV12; 350 case AV_PIX_FMT_NV16: 351 case AV_PIX_FMT_NV20: return X264_CSP_NV16; 352 }; 353 return 0; 354} 355 356#define PARSE_X264_OPT(name, var)\ 357 if (x4->var && x264_param_parse(&x4->params, name, x4->var) < 0) {\ 358 av_log(avctx, AV_LOG_ERROR, "Error parsing option '%s' with value '%s'.\n", name, x4->var);\ 359 return AVERROR(EINVAL);\ 360 } 361 362static av_cold int X264_init(AVCodecContext *avctx) 363{ 364 X264Context *x4 = avctx->priv_data; 365 int sw,sh; 366 367 if (avctx->global_quality > 0) 368 av_log(avctx, AV_LOG_WARNING, "-qscale is ignored, -crf is recommended.\n"); 369 370 x264_param_default(&x4->params); 371 372 x4->params.b_deblocking_filter = avctx->flags & CODEC_FLAG_LOOP_FILTER; 373 374 x4->params.rc.f_pb_factor = avctx->b_quant_factor; 375 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset; 376 if (x4->preset || x4->tune) 377 if (x264_param_default_preset(&x4->params, x4->preset, x4->tune) < 0) { 378 int i; 379 av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", x4->preset, x4->tune); 380 av_log(avctx, AV_LOG_INFO, "Possible presets:"); 381 for (i = 0; x264_preset_names[i]; i++) 382 av_log(avctx, AV_LOG_INFO, " %s", x264_preset_names[i]); 383 av_log(avctx, AV_LOG_INFO, "\n"); 384 av_log(avctx, AV_LOG_INFO, "Possible tunes:"); 385 for (i = 0; x264_tune_names[i]; i++) 386 av_log(avctx, AV_LOG_INFO, " %s", x264_tune_names[i]); 387 av_log(avctx, AV_LOG_INFO, "\n"); 388 return AVERROR(EINVAL); 389 } 390 391 if (avctx->level > 0) 392 x4->params.i_level_idc = avctx->level; 393 394 x4->params.pf_log = X264_log; 395 x4->params.p_log_private = avctx; 396 x4->params.i_log_level = X264_LOG_DEBUG; 397 x4->params.i_csp = convert_pix_fmt(avctx->pix_fmt); 398 399 OPT_STR("weightp", x4->wpredp); 400 401 if (avctx->bit_rate) { 402 x4->params.rc.i_bitrate = avctx->bit_rate / 1000; 403 x4->params.rc.i_rc_method = X264_RC_ABR; 404 } 405 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000; 406 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000; 407 x4->params.rc.b_stat_write = avctx->flags & CODEC_FLAG_PASS1; 408 if (avctx->flags & CODEC_FLAG_PASS2) { 409 x4->params.rc.b_stat_read = 1; 410 } else { 411 if (x4->crf >= 0) { 412 x4->params.rc.i_rc_method = X264_RC_CRF; 413 x4->params.rc.f_rf_constant = x4->crf; 414 } else if (x4->cqp >= 0) { 415 x4->params.rc.i_rc_method = X264_RC_CQP; 416 x4->params.rc.i_qp_constant = x4->cqp; 417 } 418 419 if (x4->crf_max >= 0) 420 x4->params.rc.f_rf_constant_max = x4->crf_max; 421 } 422 423 if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy > 0 && 424 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) { 425 x4->params.rc.f_vbv_buffer_init = 426 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size; 427 } 428 429 OPT_STR("level", x4->level); 430 431 if (avctx->i_quant_factor > 0) 432 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor); 433 434 if (avctx->me_method == ME_EPZS) 435 x4->params.analyse.i_me_method = X264_ME_DIA; 436 else if (avctx->me_method == ME_HEX) 437 x4->params.analyse.i_me_method = X264_ME_HEX; 438 else if (avctx->me_method == ME_UMH) 439 x4->params.analyse.i_me_method = X264_ME_UMH; 440 else if (avctx->me_method == ME_FULL) 441 x4->params.analyse.i_me_method = X264_ME_ESA; 442 else if (avctx->me_method == ME_TESA) 443 x4->params.analyse.i_me_method = X264_ME_TESA; 444 445 if (avctx->gop_size >= 0) 446 x4->params.i_keyint_max = avctx->gop_size; 447 if (avctx->max_b_frames >= 0) 448 x4->params.i_bframe = avctx->max_b_frames; 449 if (avctx->scenechange_threshold >= 0) 450 x4->params.i_scenecut_threshold = avctx->scenechange_threshold; 451 if (avctx->qmin >= 0) 452 x4->params.rc.i_qp_min = avctx->qmin; 453 if (avctx->qmax >= 0) 454 x4->params.rc.i_qp_max = avctx->qmax; 455 if (avctx->max_qdiff >= 0) 456 x4->params.rc.i_qp_step = avctx->max_qdiff; 457 if (avctx->qblur >= 0) 458 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */ 459 if (avctx->qcompress >= 0) 460 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */ 461 if (avctx->refs >= 0) 462 x4->params.i_frame_reference = avctx->refs; 463 else if (x4->level) { 464 int i; 465 int mbn = FF_CEIL_RSHIFT(avctx->width, 4) * FF_CEIL_RSHIFT(avctx->height, 4); 466 int level_id = -1; 467 char *tail; 468 int scale = X264_BUILD < 129 ? 384 : 1; 469 470 if (!strcmp(x4->level, "1b")) { 471 level_id = 9; 472 } else if (strlen(x4->level) <= 3){ 473 level_id = av_strtod(x4->level, &tail) * 10 + 0.5; 474 if (*tail) 475 level_id = -1; 476 } 477 if (level_id <= 0) 478 av_log(avctx, AV_LOG_WARNING, "Failed to parse level\n"); 479 480 for (i = 0; i<x264_levels[i].level_idc; i++) 481 if (x264_levels[i].level_idc == level_id) 482 x4->params.i_frame_reference = av_clip(x264_levels[i].dpb / mbn / scale, 1, x4->params.i_frame_reference); 483 } 484 485 if (avctx->trellis >= 0) 486 x4->params.analyse.i_trellis = avctx->trellis; 487 if (avctx->me_range >= 0) 488 x4->params.analyse.i_me_range = avctx->me_range; 489 if (avctx->noise_reduction >= 0) 490 x4->params.analyse.i_noise_reduction = avctx->noise_reduction; 491 if (avctx->me_subpel_quality >= 0) 492 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality; 493 if (avctx->b_frame_strategy >= 0) 494 x4->params.i_bframe_adaptive = avctx->b_frame_strategy; 495 if (avctx->keyint_min >= 0) 496 x4->params.i_keyint_min = avctx->keyint_min; 497 if (avctx->coder_type >= 0) 498 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC; 499 if (avctx->me_cmp >= 0) 500 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA; 501 502 if (x4->aq_mode >= 0) 503 x4->params.rc.i_aq_mode = x4->aq_mode; 504 if (x4->aq_strength >= 0) 505 x4->params.rc.f_aq_strength = x4->aq_strength; 506 PARSE_X264_OPT("psy-rd", psy_rd); 507 PARSE_X264_OPT("deblock", deblock); 508 PARSE_X264_OPT("partitions", partitions); 509 PARSE_X264_OPT("stats", stats); 510 if (x4->psy >= 0) 511 x4->params.analyse.b_psy = x4->psy; 512 if (x4->rc_lookahead >= 0) 513 x4->params.rc.i_lookahead = x4->rc_lookahead; 514 if (x4->weightp >= 0) 515 x4->params.analyse.i_weighted_pred = x4->weightp; 516 if (x4->weightb >= 0) 517 x4->params.analyse.b_weighted_bipred = x4->weightb; 518 if (x4->cplxblur >= 0) 519 x4->params.rc.f_complexity_blur = x4->cplxblur; 520 521 if (x4->ssim >= 0) 522 x4->params.analyse.b_ssim = x4->ssim; 523 if (x4->intra_refresh >= 0) 524 x4->params.b_intra_refresh = x4->intra_refresh; 525 if (x4->bluray_compat >= 0) { 526 x4->params.b_bluray_compat = x4->bluray_compat; 527 x4->params.b_vfr_input = 0; 528 } 529 if (x4->b_bias != INT_MIN) 530 x4->params.i_bframe_bias = x4->b_bias; 531 if (x4->b_pyramid >= 0) 532 x4->params.i_bframe_pyramid = x4->b_pyramid; 533 if (x4->mixed_refs >= 0) 534 x4->params.analyse.b_mixed_references = x4->mixed_refs; 535 if (x4->dct8x8 >= 0) 536 x4->params.analyse.b_transform_8x8 = x4->dct8x8; 537 if (x4->fast_pskip >= 0) 538 x4->params.analyse.b_fast_pskip = x4->fast_pskip; 539 if (x4->aud >= 0) 540 x4->params.b_aud = x4->aud; 541 if (x4->mbtree >= 0) 542 x4->params.rc.b_mb_tree = x4->mbtree; 543 if (x4->direct_pred >= 0) 544 x4->params.analyse.i_direct_mv_pred = x4->direct_pred; 545 546 if (x4->slice_max_size >= 0) 547 x4->params.i_slice_max_size = x4->slice_max_size; 548 else { 549 /* 550 * Allow x264 to be instructed through AVCodecContext about the maximum 551 * size of the RTP payload. For example, this enables the production of 552 * payload suitable for the H.264 RTP packetization-mode 0 i.e. single 553 * NAL unit per RTP packet. 554 */ 555 if (avctx->rtp_payload_size) 556 x4->params.i_slice_max_size = avctx->rtp_payload_size; 557 } 558 559 if (x4->fastfirstpass) 560 x264_param_apply_fastfirstpass(&x4->params); 561 562 /* Allow specifying the x264 profile through AVCodecContext. */ 563 if (!x4->profile) 564 switch (avctx->profile) { 565 case FF_PROFILE_H264_BASELINE: 566 x4->profile = av_strdup("baseline"); 567 break; 568 case FF_PROFILE_H264_HIGH: 569 x4->profile = av_strdup("high"); 570 break; 571 case FF_PROFILE_H264_HIGH_10: 572 x4->profile = av_strdup("high10"); 573 break; 574 case FF_PROFILE_H264_HIGH_422: 575 x4->profile = av_strdup("high422"); 576 break; 577 case FF_PROFILE_H264_HIGH_444: 578 x4->profile = av_strdup("high444"); 579 break; 580 case FF_PROFILE_H264_MAIN: 581 x4->profile = av_strdup("main"); 582 break; 583 default: 584 break; 585 } 586 587 if (x4->nal_hrd >= 0) 588 x4->params.i_nal_hrd = x4->nal_hrd; 589 590 if (x4->profile) 591 if (x264_param_apply_profile(&x4->params, x4->profile) < 0) { 592 int i; 593 av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile); 594 av_log(avctx, AV_LOG_INFO, "Possible profiles:"); 595 for (i = 0; x264_profile_names[i]; i++) 596 av_log(avctx, AV_LOG_INFO, " %s", x264_profile_names[i]); 597 av_log(avctx, AV_LOG_INFO, "\n"); 598 return AVERROR(EINVAL); 599 } 600 601 x4->params.i_width = avctx->width; 602 x4->params.i_height = avctx->height; 603 av_reduce(&sw, &sh, avctx->sample_aspect_ratio.num, avctx->sample_aspect_ratio.den, 4096); 604 x4->params.vui.i_sar_width = sw; 605 x4->params.vui.i_sar_height = sh; 606 x4->params.i_timebase_den = avctx->time_base.den; 607 x4->params.i_timebase_num = avctx->time_base.num; 608 x4->params.i_fps_num = avctx->time_base.den; 609 x4->params.i_fps_den = avctx->time_base.num * avctx->ticks_per_frame; 610 611 x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR; 612 613 x4->params.i_threads = avctx->thread_count; 614 if (avctx->thread_type) 615 x4->params.b_sliced_threads = avctx->thread_type == FF_THREAD_SLICE; 616 617 x4->params.b_interlaced = avctx->flags & CODEC_FLAG_INTERLACED_DCT; 618 619 x4->params.b_open_gop = !(avctx->flags & CODEC_FLAG_CLOSED_GOP); 620 621 x4->params.i_slice_count = avctx->slices; 622 623 x4->params.vui.b_fullrange = avctx->pix_fmt == AV_PIX_FMT_YUVJ420P || 624 avctx->pix_fmt == AV_PIX_FMT_YUVJ422P || 625 avctx->pix_fmt == AV_PIX_FMT_YUVJ444P || 626 avctx->color_range == AVCOL_RANGE_JPEG; 627 628 if (avctx->colorspace != AVCOL_SPC_UNSPECIFIED) 629 x4->params.vui.i_colmatrix = avctx->colorspace; 630 if (avctx->color_primaries != AVCOL_PRI_UNSPECIFIED) 631 x4->params.vui.i_colorprim = avctx->color_primaries; 632 if (avctx->color_trc != AVCOL_TRC_UNSPECIFIED) 633 x4->params.vui.i_transfer = avctx->color_trc; 634 635 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) 636 x4->params.b_repeat_headers = 0; 637 638 if(x4->x264opts){ 639 const char *p= x4->x264opts; 640 while(p){ 641 char param[256]={0}, val[256]={0}; 642 if(sscanf(p, "%255[^:=]=%255[^:]", param, val) == 1){ 643 OPT_STR(param, "1"); 644 }else 645 OPT_STR(param, val); 646 p= strchr(p, ':'); 647 p+=!!p; 648 } 649 } 650 651 if (x4->x264_params) { 652 AVDictionary *dict = NULL; 653 AVDictionaryEntry *en = NULL; 654 655 if (!av_dict_parse_string(&dict, x4->x264_params, "=", ":", 0)) { 656 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) { 657 if (x264_param_parse(&x4->params, en->key, en->value) < 0) 658 av_log(avctx, AV_LOG_WARNING, 659 "Error parsing option '%s = %s'.\n", 660 en->key, en->value); 661 } 662 663 av_dict_free(&dict); 664 } 665 } 666 667 // update AVCodecContext with x264 parameters 668 avctx->has_b_frames = x4->params.i_bframe ? 669 x4->params.i_bframe_pyramid ? 2 : 1 : 0; 670 if (avctx->max_b_frames < 0) 671 avctx->max_b_frames = 0; 672 673 avctx->bit_rate = x4->params.rc.i_bitrate*1000; 674 675 x4->enc = x264_encoder_open(&x4->params); 676 if (!x4->enc) 677 return -1; 678 679 avctx->coded_frame = av_frame_alloc(); 680 if (!avctx->coded_frame) 681 return AVERROR(ENOMEM); 682 683 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) { 684 x264_nal_t *nal; 685 uint8_t *p; 686 int nnal, s, i; 687 688 s = x264_encoder_headers(x4->enc, &nal, &nnal); 689 avctx->extradata = p = av_malloc(s); 690 691 for (i = 0; i < nnal; i++) { 692 /* Don't put the SEI in extradata. */ 693 if (nal[i].i_type == NAL_SEI) { 694 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25); 695 x4->sei_size = nal[i].i_payload; 696 x4->sei = av_malloc(x4->sei_size); 697 memcpy(x4->sei, nal[i].p_payload, nal[i].i_payload); 698 continue; 699 } 700 memcpy(p, nal[i].p_payload, nal[i].i_payload); 701 p += nal[i].i_payload; 702 } 703 avctx->extradata_size = p - avctx->extradata; 704 } 705 706 return 0; 707} 708 709static const enum AVPixelFormat pix_fmts_8bit[] = { 710 AV_PIX_FMT_YUV420P, 711 AV_PIX_FMT_YUVJ420P, 712 AV_PIX_FMT_YUV422P, 713 AV_PIX_FMT_YUVJ422P, 714 AV_PIX_FMT_YUV444P, 715 AV_PIX_FMT_YUVJ444P, 716 AV_PIX_FMT_NV12, 717 AV_PIX_FMT_NV16, 718 AV_PIX_FMT_NONE 719}; 720static const enum AVPixelFormat pix_fmts_9bit[] = { 721 AV_PIX_FMT_YUV420P9, 722 AV_PIX_FMT_YUV444P9, 723 AV_PIX_FMT_NONE 724}; 725static const enum AVPixelFormat pix_fmts_10bit[] = { 726 AV_PIX_FMT_YUV420P10, 727 AV_PIX_FMT_YUV422P10, 728 AV_PIX_FMT_YUV444P10, 729 AV_PIX_FMT_NV20, 730 AV_PIX_FMT_NONE 731}; 732static const enum AVPixelFormat pix_fmts_8bit_rgb[] = { 733#ifdef X264_CSP_BGR 734 AV_PIX_FMT_BGR24, 735 AV_PIX_FMT_RGB24, 736#endif 737 AV_PIX_FMT_NONE 738}; 739 740static av_cold void X264_init_static(AVCodec *codec) 741{ 742 if (x264_bit_depth == 8) 743 codec->pix_fmts = pix_fmts_8bit; 744 else if (x264_bit_depth == 9) 745 codec->pix_fmts = pix_fmts_9bit; 746 else if (x264_bit_depth == 10) 747 codec->pix_fmts = pix_fmts_10bit; 748} 749 750#define OFFSET(x) offsetof(X264Context, x) 751#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM 752static const AVOption options[] = { 753 { "preset", "Set the encoding preset (cf. x264 --fullhelp)", OFFSET(preset), AV_OPT_TYPE_STRING, { .str = "medium" }, 0, 0, VE}, 754 { "tune", "Tune the encoding params (cf. x264 --fullhelp)", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE}, 755 { "profile", "Set profile restrictions (cf. x264 --fullhelp) ", OFFSET(profile), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE}, 756 { "fastfirstpass", "Use fast settings when encoding first pass", OFFSET(fastfirstpass), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE}, 757 {"level", "Specify level (as defined by Annex A)", OFFSET(level), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE}, 758 {"passlogfile", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE}, 759 {"wpredp", "Weighted prediction for P-frames", OFFSET(wpredp), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE}, 760 {"x264opts", "x264 options", OFFSET(x264opts), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE}, 761 { "crf", "Select the quality for constant quality mode", OFFSET(crf), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE }, 762 { "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 }, 763 { "qp", "Constant quantization parameter rate control method",OFFSET(cqp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE }, 764 { "aq-mode", "AQ method", OFFSET(aq_mode), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "aq_mode"}, 765 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" }, 766 { "variance", "Variance AQ (complexity mask)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_VARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" }, 767 { "autovariance", "Auto-variance AQ (experimental)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOVARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" }, 768 { "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}, 769 { "psy", "Use psychovisual optimizations.", OFFSET(psy), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, 770 { "psy-rd", "Strength of psychovisual optimization, in <psy-rd>:<psy-trellis> format.", OFFSET(psy_rd), AV_OPT_TYPE_STRING, {0 }, 0, 0, VE}, 771 { "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 }, 772 { "weightb", "Weighted prediction for B-frames.", OFFSET(weightb), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, 773 { "weightp", "Weighted prediction analysis method.", OFFSET(weightp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "weightp" }, 774 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_NONE}, INT_MIN, INT_MAX, VE, "weightp" }, 775 { "simple", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SIMPLE}, INT_MIN, INT_MAX, VE, "weightp" }, 776 { "smart", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_WEIGHTP_SMART}, INT_MIN, INT_MAX, VE, "weightp" }, 777 { "ssim", "Calculate and print SSIM stats.", OFFSET(ssim), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, 778 { "intra-refresh", "Use Periodic Intra Refresh instead of IDR frames.",OFFSET(intra_refresh),AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, 779 { "bluray-compat", "Bluray compatibility workarounds.", OFFSET(bluray_compat) ,AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, 780 { "b-bias", "Influences how often B-frames are used", OFFSET(b_bias), AV_OPT_TYPE_INT, { .i64 = INT_MIN}, INT_MIN, INT_MAX, VE }, 781 { "b-pyramid", "Keep some B-frames as references.", OFFSET(b_pyramid), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "b_pyramid" }, 782 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NONE}, INT_MIN, INT_MAX, VE, "b_pyramid" }, 783 { "strict", "Strictly hierarchical pyramid", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_STRICT}, INT_MIN, INT_MAX, VE, "b_pyramid" }, 784 { "normal", "Non-strict (not Blu-ray compatible)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_B_PYRAMID_NORMAL}, INT_MIN, INT_MAX, VE, "b_pyramid" }, 785 { "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 }, 786 { "8x8dct", "High profile 8x8 transform.", OFFSET(dct8x8), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE}, 787 { "fast-pskip", NULL, OFFSET(fast_pskip), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE}, 788 { "aud", "Use access unit delimiters.", OFFSET(aud), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE}, 789 { "mbtree", "Use macroblock tree ratecontrol.", OFFSET(mbtree), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE}, 790 { "deblock", "Loop filter parameters, in <alpha:beta> form.", OFFSET(deblock), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE}, 791 { "cplxblur", "Reduce fluctuations in QP (before curve compression)", OFFSET(cplxblur), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE}, 792 { "partitions", "A comma-separated list of partitions to consider. " 793 "Possible values: p8x8, p4x4, b8x8, i8x8, i4x4, none, all", OFFSET(partitions), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE}, 794 { "direct-pred", "Direct MV prediction mode", OFFSET(direct_pred), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "direct-pred" }, 795 { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_NONE }, 0, 0, VE, "direct-pred" }, 796 { "spatial", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_SPATIAL }, 0, 0, VE, "direct-pred" }, 797 { "temporal", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_TEMPORAL }, 0, 0, VE, "direct-pred" }, 798 { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = X264_DIRECT_PRED_AUTO }, 0, 0, VE, "direct-pred" }, 799 { "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 }, 800 { "stats", "Filename for 2 pass stats", OFFSET(stats), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE }, 801 { "nal-hrd", "Signal HRD information (requires vbv-bufsize; " 802 "cbr not allowed in .mp4)", OFFSET(nal_hrd), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "nal-hrd" }, 803 { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_NONE}, INT_MIN, INT_MAX, VE, "nal-hrd" }, 804 { "vbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_VBR}, INT_MIN, INT_MAX, VE, "nal-hrd" }, 805 { "cbr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_NAL_HRD_CBR}, INT_MIN, INT_MAX, VE, "nal-hrd" }, 806 { "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 }, 807 { NULL }, 808}; 809 810static const AVClass x264_class = { 811 .class_name = "libx264", 812 .item_name = av_default_item_name, 813 .option = options, 814 .version = LIBAVUTIL_VERSION_INT, 815}; 816 817static const AVClass rgbclass = { 818 .class_name = "libx264rgb", 819 .item_name = av_default_item_name, 820 .option = options, 821 .version = LIBAVUTIL_VERSION_INT, 822}; 823 824static const AVCodecDefault x264_defaults[] = { 825 { "b", "0" }, 826 { "bf", "-1" }, 827 { "flags2", "0" }, 828 { "g", "-1" }, 829 { "i_qfactor", "-1" }, 830 { "qmin", "-1" }, 831 { "qmax", "-1" }, 832 { "qdiff", "-1" }, 833 { "qblur", "-1" }, 834 { "qcomp", "-1" }, 835// { "rc_lookahead", "-1" }, 836 { "refs", "-1" }, 837 { "sc_threshold", "-1" }, 838 { "trellis", "-1" }, 839 { "nr", "-1" }, 840 { "me_range", "-1" }, 841 { "me_method", "-1" }, 842 { "subq", "-1" }, 843 { "b_strategy", "-1" }, 844 { "keyint_min", "-1" }, 845 { "coder", "-1" }, 846 { "cmp", "-1" }, 847 { "threads", AV_STRINGIFY(X264_THREADS_AUTO) }, 848 { "thread_type", "0" }, 849 { "flags", "+cgop" }, 850 { "rc_init_occupancy","-1" }, 851 { NULL }, 852}; 853 854AVCodec ff_libx264_encoder = { 855 .name = "libx264", 856 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"), 857 .type = AVMEDIA_TYPE_VIDEO, 858 .id = AV_CODEC_ID_H264, 859 .priv_data_size = sizeof(X264Context), 860 .init = X264_init, 861 .encode2 = X264_frame, 862 .close = X264_close, 863 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS, 864 .priv_class = &x264_class, 865 .defaults = x264_defaults, 866 .init_static_data = X264_init_static, 867}; 868 869AVCodec ff_libx264rgb_encoder = { 870 .name = "libx264rgb", 871 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 RGB"), 872 .type = AVMEDIA_TYPE_VIDEO, 873 .id = AV_CODEC_ID_H264, 874 .priv_data_size = sizeof(X264Context), 875 .init = X264_init, 876 .encode2 = X264_frame, 877 .close = X264_close, 878 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS, 879 .priv_class = &rgbclass, 880 .defaults = x264_defaults, 881 .pix_fmts = pix_fmts_8bit_rgb, 882}; 883