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.

821 lines
32KB

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