You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

778 lines
30KB

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