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.

616 lines
25KB

  1. /*
  2. * Copyright (c) 2010, Google, Inc.
  3. *
  4. * This file is part of FFmpeg.
  5. *
  6. * FFmpeg is free software; you can redistribute it and/or
  7. * modify it under the terms of the GNU Lesser General Public
  8. * License as published by the Free Software Foundation; either
  9. * version 2.1 of the License, or (at your option) any later version.
  10. *
  11. * FFmpeg is distributed in the hope that it will be useful,
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14. * Lesser General Public License for more details.
  15. *
  16. * You should have received a copy of the GNU Lesser General Public
  17. * License along with FFmpeg; if not, write to the Free Software
  18. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19. */
  20. /**
  21. * @file
  22. * VP8 encoder support via libvpx
  23. */
  24. #define VPX_DISABLE_CTRL_TYPECHECKS 1
  25. #define VPX_CODEC_DISABLE_COMPAT 1
  26. #include <vpx/vpx_encoder.h>
  27. #include <vpx/vp8cx.h>
  28. #include "avcodec.h"
  29. #include "internal.h"
  30. #include "libavutil/avassert.h"
  31. #include "libavutil/base64.h"
  32. #include "libavutil/mathematics.h"
  33. #include "libavutil/opt.h"
  34. /**
  35. * Portion of struct vpx_codec_cx_pkt from vpx_encoder.h.
  36. * One encoded frame returned from the library.
  37. */
  38. struct FrameListData {
  39. void *buf; /**< compressed data buffer */
  40. size_t sz; /**< length of compressed data */
  41. int64_t pts; /**< time stamp to show frame
  42. (in timebase units) */
  43. unsigned long duration; /**< duration to show frame
  44. (in timebase units) */
  45. uint32_t flags; /**< flags for this frame */
  46. struct FrameListData *next;
  47. };
  48. typedef struct VP8EncoderContext {
  49. AVClass *class;
  50. struct vpx_codec_ctx encoder;
  51. struct vpx_image rawimg;
  52. struct vpx_fixed_buf twopass_stats;
  53. int deadline; //i.e., RT/GOOD/BEST
  54. struct FrameListData *coded_frame_list;
  55. int cpu_used;
  56. /**
  57. * VP8 specific flags, see VP8F_* below.
  58. */
  59. int flags;
  60. #define VP8F_ERROR_RESILIENT 0x00000001 ///< Enable measures appropriate for streaming over lossy links
  61. #define VP8F_AUTO_ALT_REF 0x00000002 ///< Enable automatic alternate reference frame generation
  62. int auto_alt_ref;
  63. int arnr_max_frames;
  64. int arnr_strength;
  65. int arnr_type;
  66. int lag_in_frames;
  67. int error_resilient;
  68. int crf;
  69. int max_intra_rate;
  70. } VP8Context;
  71. /** String mappings for enum vp8e_enc_control_id */
  72. static const char *const ctlidstr[] = {
  73. [VP8E_UPD_ENTROPY] = "VP8E_UPD_ENTROPY",
  74. [VP8E_UPD_REFERENCE] = "VP8E_UPD_REFERENCE",
  75. [VP8E_USE_REFERENCE] = "VP8E_USE_REFERENCE",
  76. [VP8E_SET_ROI_MAP] = "VP8E_SET_ROI_MAP",
  77. [VP8E_SET_ACTIVEMAP] = "VP8E_SET_ACTIVEMAP",
  78. [VP8E_SET_SCALEMODE] = "VP8E_SET_SCALEMODE",
  79. [VP8E_SET_CPUUSED] = "VP8E_SET_CPUUSED",
  80. [VP8E_SET_ENABLEAUTOALTREF] = "VP8E_SET_ENABLEAUTOALTREF",
  81. [VP8E_SET_NOISE_SENSITIVITY] = "VP8E_SET_NOISE_SENSITIVITY",
  82. [VP8E_SET_SHARPNESS] = "VP8E_SET_SHARPNESS",
  83. [VP8E_SET_STATIC_THRESHOLD] = "VP8E_SET_STATIC_THRESHOLD",
  84. [VP8E_SET_TOKEN_PARTITIONS] = "VP8E_SET_TOKEN_PARTITIONS",
  85. [VP8E_GET_LAST_QUANTIZER] = "VP8E_GET_LAST_QUANTIZER",
  86. [VP8E_SET_ARNR_MAXFRAMES] = "VP8E_SET_ARNR_MAXFRAMES",
  87. [VP8E_SET_ARNR_STRENGTH] = "VP8E_SET_ARNR_STRENGTH",
  88. [VP8E_SET_ARNR_TYPE] = "VP8E_SET_ARNR_TYPE",
  89. [VP8E_SET_CQ_LEVEL] = "VP8E_SET_CQ_LEVEL",
  90. [VP8E_SET_MAX_INTRA_BITRATE_PCT] = "VP8E_SET_MAX_INTRA_BITRATE_PCT",
  91. };
  92. static av_cold void log_encoder_error(AVCodecContext *avctx, const char *desc)
  93. {
  94. VP8Context *ctx = avctx->priv_data;
  95. const char *error = vpx_codec_error(&ctx->encoder);
  96. const char *detail = vpx_codec_error_detail(&ctx->encoder);
  97. av_log(avctx, AV_LOG_ERROR, "%s: %s\n", desc, error);
  98. if (detail)
  99. av_log(avctx, AV_LOG_ERROR, " Additional information: %s\n", detail);
  100. }
  101. static av_cold void dump_enc_cfg(AVCodecContext *avctx,
  102. const struct vpx_codec_enc_cfg *cfg)
  103. {
  104. int width = -30;
  105. int level = AV_LOG_DEBUG;
  106. av_log(avctx, level, "vpx_codec_enc_cfg\n");
  107. av_log(avctx, level, "generic settings\n"
  108. " %*s%u\n %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
  109. " %*s{%u/%u}\n %*s%u\n %*s%d\n %*s%u\n",
  110. width, "g_usage:", cfg->g_usage,
  111. width, "g_threads:", cfg->g_threads,
  112. width, "g_profile:", cfg->g_profile,
  113. width, "g_w:", cfg->g_w,
  114. width, "g_h:", cfg->g_h,
  115. width, "g_timebase:", cfg->g_timebase.num, cfg->g_timebase.den,
  116. width, "g_error_resilient:", cfg->g_error_resilient,
  117. width, "g_pass:", cfg->g_pass,
  118. width, "g_lag_in_frames:", cfg->g_lag_in_frames);
  119. av_log(avctx, level, "rate control settings\n"
  120. " %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
  121. " %*s%d\n %*s%p(%zu)\n %*s%u\n",
  122. width, "rc_dropframe_thresh:", cfg->rc_dropframe_thresh,
  123. width, "rc_resize_allowed:", cfg->rc_resize_allowed,
  124. width, "rc_resize_up_thresh:", cfg->rc_resize_up_thresh,
  125. width, "rc_resize_down_thresh:", cfg->rc_resize_down_thresh,
  126. width, "rc_end_usage:", cfg->rc_end_usage,
  127. width, "rc_twopass_stats_in:", cfg->rc_twopass_stats_in.buf, cfg->rc_twopass_stats_in.sz,
  128. width, "rc_target_bitrate:", cfg->rc_target_bitrate);
  129. av_log(avctx, level, "quantizer settings\n"
  130. " %*s%u\n %*s%u\n",
  131. width, "rc_min_quantizer:", cfg->rc_min_quantizer,
  132. width, "rc_max_quantizer:", cfg->rc_max_quantizer);
  133. av_log(avctx, level, "bitrate tolerance\n"
  134. " %*s%u\n %*s%u\n",
  135. width, "rc_undershoot_pct:", cfg->rc_undershoot_pct,
  136. width, "rc_overshoot_pct:", cfg->rc_overshoot_pct);
  137. av_log(avctx, level, "decoder buffer model\n"
  138. " %*s%u\n %*s%u\n %*s%u\n",
  139. width, "rc_buf_sz:", cfg->rc_buf_sz,
  140. width, "rc_buf_initial_sz:", cfg->rc_buf_initial_sz,
  141. width, "rc_buf_optimal_sz:", cfg->rc_buf_optimal_sz);
  142. av_log(avctx, level, "2 pass rate control settings\n"
  143. " %*s%u\n %*s%u\n %*s%u\n",
  144. width, "rc_2pass_vbr_bias_pct:", cfg->rc_2pass_vbr_bias_pct,
  145. width, "rc_2pass_vbr_minsection_pct:", cfg->rc_2pass_vbr_minsection_pct,
  146. width, "rc_2pass_vbr_maxsection_pct:", cfg->rc_2pass_vbr_maxsection_pct);
  147. av_log(avctx, level, "keyframing settings\n"
  148. " %*s%d\n %*s%u\n %*s%u\n",
  149. width, "kf_mode:", cfg->kf_mode,
  150. width, "kf_min_dist:", cfg->kf_min_dist,
  151. width, "kf_max_dist:", cfg->kf_max_dist);
  152. av_log(avctx, level, "\n");
  153. }
  154. static void coded_frame_add(void *list, struct FrameListData *cx_frame)
  155. {
  156. struct FrameListData **p = list;
  157. while (*p != NULL)
  158. p = &(*p)->next;
  159. *p = cx_frame;
  160. cx_frame->next = NULL;
  161. }
  162. static av_cold void free_coded_frame(struct FrameListData *cx_frame)
  163. {
  164. av_freep(&cx_frame->buf);
  165. av_freep(&cx_frame);
  166. }
  167. static av_cold void free_frame_list(struct FrameListData *list)
  168. {
  169. struct FrameListData *p = list;
  170. while (p) {
  171. list = list->next;
  172. free_coded_frame(p);
  173. p = list;
  174. }
  175. }
  176. static av_cold int codecctl_int(AVCodecContext *avctx,
  177. enum vp8e_enc_control_id id, int val)
  178. {
  179. VP8Context *ctx = avctx->priv_data;
  180. char buf[80];
  181. int width = -30;
  182. int res;
  183. snprintf(buf, sizeof(buf), "%s:", ctlidstr[id]);
  184. av_log(avctx, AV_LOG_DEBUG, " %*s%d\n", width, buf, val);
  185. res = vpx_codec_control(&ctx->encoder, id, val);
  186. if (res != VPX_CODEC_OK) {
  187. snprintf(buf, sizeof(buf), "Failed to set %s codec control",
  188. ctlidstr[id]);
  189. log_encoder_error(avctx, buf);
  190. }
  191. return res == VPX_CODEC_OK ? 0 : AVERROR(EINVAL);
  192. }
  193. static av_cold int vp8_free(AVCodecContext *avctx)
  194. {
  195. VP8Context *ctx = avctx->priv_data;
  196. vpx_codec_destroy(&ctx->encoder);
  197. av_freep(&ctx->twopass_stats.buf);
  198. av_freep(&avctx->coded_frame);
  199. av_freep(&avctx->stats_out);
  200. free_frame_list(ctx->coded_frame_list);
  201. return 0;
  202. }
  203. static av_cold int vp8_init(AVCodecContext *avctx)
  204. {
  205. VP8Context *ctx = avctx->priv_data;
  206. const struct vpx_codec_iface *iface = &vpx_codec_vp8_cx_algo;
  207. struct vpx_codec_enc_cfg enccfg;
  208. int res;
  209. av_log(avctx, AV_LOG_INFO, "%s\n", vpx_codec_version_str());
  210. av_log(avctx, AV_LOG_VERBOSE, "%s\n", vpx_codec_build_config());
  211. if ((res = vpx_codec_enc_config_default(iface, &enccfg, 0)) != VPX_CODEC_OK) {
  212. av_log(avctx, AV_LOG_ERROR, "Failed to get config: %s\n",
  213. vpx_codec_err_to_string(res));
  214. return AVERROR(EINVAL);
  215. }
  216. dump_enc_cfg(avctx, &enccfg);
  217. enccfg.g_w = avctx->width;
  218. enccfg.g_h = avctx->height;
  219. enccfg.g_timebase.num = avctx->time_base.num;
  220. enccfg.g_timebase.den = avctx->time_base.den;
  221. enccfg.g_threads = avctx->thread_count;
  222. enccfg.g_lag_in_frames= ctx->lag_in_frames;
  223. if (avctx->flags & CODEC_FLAG_PASS1)
  224. enccfg.g_pass = VPX_RC_FIRST_PASS;
  225. else if (avctx->flags & CODEC_FLAG_PASS2)
  226. enccfg.g_pass = VPX_RC_LAST_PASS;
  227. else
  228. enccfg.g_pass = VPX_RC_ONE_PASS;
  229. if (avctx->rc_min_rate == avctx->rc_max_rate &&
  230. avctx->rc_min_rate == avctx->bit_rate)
  231. enccfg.rc_end_usage = VPX_CBR;
  232. else if (ctx->crf)
  233. enccfg.rc_end_usage = VPX_CQ;
  234. enccfg.rc_target_bitrate = av_rescale_rnd(avctx->bit_rate, 1, 1000,
  235. AV_ROUND_NEAR_INF);
  236. if (avctx->qmin > 0)
  237. enccfg.rc_min_quantizer = avctx->qmin;
  238. if (avctx->qmax > 0)
  239. enccfg.rc_max_quantizer = avctx->qmax;
  240. enccfg.rc_dropframe_thresh = avctx->frame_skip_threshold;
  241. //0-100 (0 => CBR, 100 => VBR)
  242. enccfg.rc_2pass_vbr_bias_pct = round(avctx->qcompress * 100);
  243. enccfg.rc_2pass_vbr_minsection_pct =
  244. avctx->rc_min_rate * 100LL / avctx->bit_rate;
  245. if (avctx->rc_max_rate)
  246. enccfg.rc_2pass_vbr_maxsection_pct =
  247. avctx->rc_max_rate * 100LL / avctx->bit_rate;
  248. if (avctx->rc_buffer_size)
  249. enccfg.rc_buf_sz =
  250. avctx->rc_buffer_size * 1000LL / avctx->bit_rate;
  251. if (avctx->rc_initial_buffer_occupancy)
  252. enccfg.rc_buf_initial_sz =
  253. avctx->rc_initial_buffer_occupancy * 1000LL / avctx->bit_rate;
  254. enccfg.rc_buf_optimal_sz = enccfg.rc_buf_sz * 5 / 6;
  255. enccfg.rc_undershoot_pct = round(avctx->rc_buffer_aggressivity * 100);
  256. //_enc_init() will balk if kf_min_dist differs from max w/VPX_KF_AUTO
  257. if (avctx->keyint_min >= 0 && avctx->keyint_min == avctx->gop_size)
  258. enccfg.kf_min_dist = avctx->keyint_min;
  259. if (avctx->gop_size >= 0)
  260. enccfg.kf_max_dist = avctx->gop_size;
  261. if (enccfg.g_pass == VPX_RC_FIRST_PASS)
  262. enccfg.g_lag_in_frames = 0;
  263. else if (enccfg.g_pass == VPX_RC_LAST_PASS) {
  264. int decode_size;
  265. if (!avctx->stats_in) {
  266. av_log(avctx, AV_LOG_ERROR, "No stats file for second pass\n");
  267. return AVERROR_INVALIDDATA;
  268. }
  269. ctx->twopass_stats.sz = strlen(avctx->stats_in) * 3 / 4;
  270. ctx->twopass_stats.buf = av_malloc(ctx->twopass_stats.sz);
  271. if (!ctx->twopass_stats.buf) {
  272. av_log(avctx, AV_LOG_ERROR,
  273. "Stat buffer alloc (%zu bytes) failed\n",
  274. ctx->twopass_stats.sz);
  275. return AVERROR(ENOMEM);
  276. }
  277. decode_size = av_base64_decode(ctx->twopass_stats.buf, avctx->stats_in,
  278. ctx->twopass_stats.sz);
  279. if (decode_size < 0) {
  280. av_log(avctx, AV_LOG_ERROR, "Stat buffer decode failed\n");
  281. return AVERROR_INVALIDDATA;
  282. }
  283. ctx->twopass_stats.sz = decode_size;
  284. enccfg.rc_twopass_stats_in = ctx->twopass_stats;
  285. }
  286. /* 0-3: For non-zero values the encoder increasingly optimizes for reduced
  287. complexity playback on low powered devices at the expense of encode
  288. quality. */
  289. if (avctx->profile != FF_PROFILE_UNKNOWN)
  290. enccfg.g_profile = avctx->profile;
  291. enccfg.g_error_resilient = ctx->error_resilient || ctx->flags & VP8F_ERROR_RESILIENT;
  292. dump_enc_cfg(avctx, &enccfg);
  293. /* Construct Encoder Context */
  294. res = vpx_codec_enc_init(&ctx->encoder, iface, &enccfg, 0);
  295. if (res != VPX_CODEC_OK) {
  296. log_encoder_error(avctx, "Failed to initialize encoder");
  297. return AVERROR(EINVAL);
  298. }
  299. //codec control failures are currently treated only as warnings
  300. av_log(avctx, AV_LOG_DEBUG, "vpx_codec_control\n");
  301. if (ctx->cpu_used != INT_MIN)
  302. codecctl_int(avctx, VP8E_SET_CPUUSED, ctx->cpu_used);
  303. if (ctx->flags & VP8F_AUTO_ALT_REF)
  304. ctx->auto_alt_ref = 1;
  305. if (ctx->auto_alt_ref >= 0)
  306. codecctl_int(avctx, VP8E_SET_ENABLEAUTOALTREF, ctx->auto_alt_ref);
  307. if (ctx->arnr_max_frames >= 0)
  308. codecctl_int(avctx, VP8E_SET_ARNR_MAXFRAMES, ctx->arnr_max_frames);
  309. if (ctx->arnr_strength >= 0)
  310. codecctl_int(avctx, VP8E_SET_ARNR_STRENGTH, ctx->arnr_strength);
  311. if (ctx->arnr_type >= 0)
  312. codecctl_int(avctx, VP8E_SET_ARNR_TYPE, ctx->arnr_type);
  313. codecctl_int(avctx, VP8E_SET_NOISE_SENSITIVITY, avctx->noise_reduction);
  314. codecctl_int(avctx, VP8E_SET_TOKEN_PARTITIONS, av_log2(avctx->slices));
  315. codecctl_int(avctx, VP8E_SET_STATIC_THRESHOLD, avctx->mb_threshold);
  316. codecctl_int(avctx, VP8E_SET_CQ_LEVEL, ctx->crf);
  317. if (ctx->max_intra_rate >= 0)
  318. codecctl_int(avctx, VP8E_SET_MAX_INTRA_BITRATE_PCT, ctx->max_intra_rate);
  319. av_log(avctx, AV_LOG_DEBUG, "Using deadline: %d\n", ctx->deadline);
  320. //provide dummy value to initialize wrapper, values will be updated each _encode()
  321. vpx_img_wrap(&ctx->rawimg, VPX_IMG_FMT_I420, avctx->width, avctx->height, 1,
  322. (unsigned char*)1);
  323. avctx->coded_frame = avcodec_alloc_frame();
  324. if (!avctx->coded_frame) {
  325. av_log(avctx, AV_LOG_ERROR, "Error allocating coded frame\n");
  326. vp8_free(avctx);
  327. return AVERROR(ENOMEM);
  328. }
  329. return 0;
  330. }
  331. static inline void cx_pktcpy(struct FrameListData *dst,
  332. const struct vpx_codec_cx_pkt *src)
  333. {
  334. dst->pts = src->data.frame.pts;
  335. dst->duration = src->data.frame.duration;
  336. dst->flags = src->data.frame.flags;
  337. dst->sz = src->data.frame.sz;
  338. dst->buf = src->data.frame.buf;
  339. }
  340. /**
  341. * Store coded frame information in format suitable for return from encode2().
  342. *
  343. * Write information from @a cx_frame to @a pkt
  344. * @return packet data size on success
  345. * @return a negative AVERROR on error
  346. */
  347. static int storeframe(AVCodecContext *avctx, struct FrameListData *cx_frame,
  348. AVPacket *pkt, AVFrame *coded_frame)
  349. {
  350. int ret = ff_alloc_packet2(avctx, pkt, cx_frame->sz);
  351. if (ret >= 0) {
  352. memcpy(pkt->data, cx_frame->buf, pkt->size);
  353. pkt->pts = pkt->dts = cx_frame->pts;
  354. coded_frame->pts = cx_frame->pts;
  355. coded_frame->key_frame = !!(cx_frame->flags & VPX_FRAME_IS_KEY);
  356. if (coded_frame->key_frame) {
  357. coded_frame->pict_type = AV_PICTURE_TYPE_I;
  358. pkt->flags |= AV_PKT_FLAG_KEY;
  359. } else
  360. coded_frame->pict_type = AV_PICTURE_TYPE_P;
  361. } else {
  362. return ret;
  363. }
  364. return pkt->size;
  365. }
  366. /**
  367. * Queue multiple output frames from the encoder, returning the front-most.
  368. * In cases where vpx_codec_get_cx_data() returns more than 1 frame append
  369. * the frame queue. Return the head frame if available.
  370. * @return Stored frame size
  371. * @return AVERROR(EINVAL) on output size error
  372. * @return AVERROR(ENOMEM) on coded frame queue data allocation error
  373. */
  374. static int queue_frames(AVCodecContext *avctx, AVPacket *pkt_out,
  375. AVFrame *coded_frame)
  376. {
  377. VP8Context *ctx = avctx->priv_data;
  378. const struct vpx_codec_cx_pkt *pkt;
  379. const void *iter = NULL;
  380. int size = 0;
  381. if (ctx->coded_frame_list) {
  382. struct FrameListData *cx_frame = ctx->coded_frame_list;
  383. /* return the leading frame if we've already begun queueing */
  384. size = storeframe(avctx, cx_frame, pkt_out, coded_frame);
  385. if (size < 0)
  386. return size;
  387. ctx->coded_frame_list = cx_frame->next;
  388. free_coded_frame(cx_frame);
  389. }
  390. /* consume all available output from the encoder before returning. buffers
  391. are only good through the next vpx_codec call */
  392. while ((pkt = vpx_codec_get_cx_data(&ctx->encoder, &iter))) {
  393. switch (pkt->kind) {
  394. case VPX_CODEC_CX_FRAME_PKT:
  395. if (!size) {
  396. struct FrameListData cx_frame;
  397. /* avoid storing the frame when the list is empty and we haven't yet
  398. provided a frame for output */
  399. av_assert0(!ctx->coded_frame_list);
  400. cx_pktcpy(&cx_frame, pkt);
  401. size = storeframe(avctx, &cx_frame, pkt_out, coded_frame);
  402. if (size < 0)
  403. return size;
  404. } else {
  405. struct FrameListData *cx_frame =
  406. av_malloc(sizeof(struct FrameListData));
  407. if (!cx_frame) {
  408. av_log(avctx, AV_LOG_ERROR,
  409. "Frame queue element alloc failed\n");
  410. return AVERROR(ENOMEM);
  411. }
  412. cx_pktcpy(cx_frame, pkt);
  413. cx_frame->buf = av_malloc(cx_frame->sz);
  414. if (!cx_frame->buf) {
  415. av_log(avctx, AV_LOG_ERROR,
  416. "Data buffer alloc (%zu bytes) failed\n",
  417. cx_frame->sz);
  418. return AVERROR(ENOMEM);
  419. }
  420. memcpy(cx_frame->buf, pkt->data.frame.buf, pkt->data.frame.sz);
  421. coded_frame_add(&ctx->coded_frame_list, cx_frame);
  422. }
  423. break;
  424. case VPX_CODEC_STATS_PKT: {
  425. struct vpx_fixed_buf *stats = &ctx->twopass_stats;
  426. stats->buf = av_realloc_f(stats->buf, 1,
  427. stats->sz + pkt->data.twopass_stats.sz);
  428. if (!stats->buf) {
  429. av_log(avctx, AV_LOG_ERROR, "Stat buffer realloc failed\n");
  430. return AVERROR(ENOMEM);
  431. }
  432. memcpy((uint8_t*)stats->buf + stats->sz,
  433. pkt->data.twopass_stats.buf, pkt->data.twopass_stats.sz);
  434. stats->sz += pkt->data.twopass_stats.sz;
  435. break;
  436. }
  437. case VPX_CODEC_PSNR_PKT: //FIXME add support for CODEC_FLAG_PSNR
  438. case VPX_CODEC_CUSTOM_PKT:
  439. //ignore unsupported/unrecognized packet types
  440. break;
  441. }
  442. }
  443. return size;
  444. }
  445. static int vp8_encode(AVCodecContext *avctx, AVPacket *pkt,
  446. const AVFrame *frame, int *got_packet)
  447. {
  448. VP8Context *ctx = avctx->priv_data;
  449. struct vpx_image *rawimg = NULL;
  450. int64_t timestamp = 0;
  451. long flags = 0;
  452. int res, coded_size;
  453. if (frame) {
  454. rawimg = &ctx->rawimg;
  455. rawimg->planes[VPX_PLANE_Y] = frame->data[0];
  456. rawimg->planes[VPX_PLANE_U] = frame->data[1];
  457. rawimg->planes[VPX_PLANE_V] = frame->data[2];
  458. rawimg->stride[VPX_PLANE_Y] = frame->linesize[0];
  459. rawimg->stride[VPX_PLANE_U] = frame->linesize[1];
  460. rawimg->stride[VPX_PLANE_V] = frame->linesize[2];
  461. timestamp = frame->pts;
  462. flags = frame->pict_type == AV_PICTURE_TYPE_I ? VPX_EFLAG_FORCE_KF : 0;
  463. }
  464. res = vpx_codec_encode(&ctx->encoder, rawimg, timestamp,
  465. avctx->ticks_per_frame, flags, ctx->deadline);
  466. if (res != VPX_CODEC_OK) {
  467. log_encoder_error(avctx, "Error encoding frame");
  468. return AVERROR_INVALIDDATA;
  469. }
  470. coded_size = queue_frames(avctx, pkt, avctx->coded_frame);
  471. if (!frame && avctx->flags & CODEC_FLAG_PASS1) {
  472. unsigned int b64_size = AV_BASE64_SIZE(ctx->twopass_stats.sz);
  473. avctx->stats_out = av_malloc(b64_size);
  474. if (!avctx->stats_out) {
  475. av_log(avctx, AV_LOG_ERROR, "Stat buffer alloc (%d bytes) failed\n",
  476. b64_size);
  477. return AVERROR(ENOMEM);
  478. }
  479. av_base64_encode(avctx->stats_out, b64_size, ctx->twopass_stats.buf,
  480. ctx->twopass_stats.sz);
  481. }
  482. *got_packet = !!coded_size;
  483. return 0;
  484. }
  485. #define OFFSET(x) offsetof(VP8Context, x)
  486. #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
  487. static const AVOption options[] = {
  488. { "cpu-used", "Quality/Speed ratio modifier", OFFSET(cpu_used), AV_OPT_TYPE_INT, {INT_MIN}, INT_MIN, INT_MAX, VE},
  489. { "auto-alt-ref", "Enable use of alternate reference "
  490. "frames (2-pass only)", OFFSET(auto_alt_ref), AV_OPT_TYPE_INT, {-1}, -1, 1, VE},
  491. { "lag-in-frames", "Number of frames to look ahead for "
  492. "alternate reference frame selection", OFFSET(lag_in_frames), AV_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  493. { "arnr-maxframes", "altref noise reduction max frame count", OFFSET(arnr_max_frames), AV_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  494. { "arnr-strength", "altref noise reduction filter strength", OFFSET(arnr_strength), AV_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  495. { "arnr-type", "altref noise reduction filter type", OFFSET(arnr_type), AV_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE, "arnr_type"},
  496. { "backward", NULL, 0, AV_OPT_TYPE_CONST, {1}, 0, 0, VE, "arnr_type" },
  497. { "forward", NULL, 0, AV_OPT_TYPE_CONST, {2}, 0, 0, VE, "arnr_type" },
  498. { "centered", NULL, 0, AV_OPT_TYPE_CONST, {3}, 0, 0, VE, "arnr_type" },
  499. { "deadline", "Time to spend encoding, in microseconds.", OFFSET(deadline), AV_OPT_TYPE_INT, {VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  500. { "best", NULL, 0, AV_OPT_TYPE_CONST, {VPX_DL_BEST_QUALITY}, 0, 0, VE, "quality"},
  501. { "good", NULL, 0, AV_OPT_TYPE_CONST, {VPX_DL_GOOD_QUALITY}, 0, 0, VE, "quality"},
  502. { "realtime", NULL, 0, AV_OPT_TYPE_CONST, {VPX_DL_REALTIME}, 0, 0, VE, "quality"},
  503. { "error-resilient", "Error resilience configuration", OFFSET(error_resilient), AV_OPT_TYPE_FLAGS, {0}, INT_MIN, INT_MAX, VE, "er"},
  504. { "max-intra-rate", "Maximum I-frame bitrate (pct) 0=unlimited", OFFSET(max_intra_rate), AV_OPT_TYPE_INT, {-1}, -1, INT_MAX, VE},
  505. #ifdef VPX_ERROR_RESILIENT_DEFAULT
  506. { "default", "Improve resiliency against losses of whole frames", 0, AV_OPT_TYPE_CONST, {VPX_ERROR_RESILIENT_DEFAULT}, 0, 0, VE, "er"},
  507. { "partitions", "The frame partitions are independently decodable "
  508. "by the bool decoder, meaning that partitions can be decoded even "
  509. "though earlier partitions have been lost. Note that intra predicition"
  510. " is still done over the partition boundary.", 0, AV_OPT_TYPE_CONST, {VPX_ERROR_RESILIENT_PARTITIONS}, 0, 0, VE, "er"},
  511. #endif
  512. {"speed", "", offsetof(VP8Context, cpu_used), AV_OPT_TYPE_INT, {.dbl = 3}, -16, 16, VE},
  513. {"quality", "", offsetof(VP8Context, deadline), AV_OPT_TYPE_INT, {.dbl = VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"},
  514. {"vp8flags", "", offsetof(VP8Context, flags), FF_OPT_TYPE_FLAGS, {.dbl = 0}, 0, UINT_MAX, VE, "flags"},
  515. {"error_resilient", "enable error resilience", 0, FF_OPT_TYPE_CONST, {.dbl = VP8F_ERROR_RESILIENT}, INT_MIN, INT_MAX, VE, "flags"},
  516. {"altref", "enable use of alternate reference frames (VP8/2-pass only)", 0, FF_OPT_TYPE_CONST, {.dbl = VP8F_AUTO_ALT_REF}, INT_MIN, INT_MAX, VE, "flags"},
  517. {"arnr_max_frames", "altref noise reduction max frame count", offsetof(VP8Context, arnr_max_frames), AV_OPT_TYPE_INT, {.dbl = 0}, 0, 15, VE},
  518. {"arnr_strength", "altref noise reduction filter strength", offsetof(VP8Context, arnr_strength), AV_OPT_TYPE_INT, {.dbl = 3}, 0, 6, VE},
  519. {"arnr_type", "altref noise reduction filter type", offsetof(VP8Context, arnr_type), AV_OPT_TYPE_INT, {.dbl = 3}, 1, 3, VE},
  520. {"rc_lookahead", "Number of frames to look ahead for alternate reference frame selection", offsetof(VP8Context, lag_in_frames), AV_OPT_TYPE_INT, {.dbl = 25}, 0, 25, VE},
  521. {"crf", "Select the quality for constant quality mode", offsetof(VP8Context, crf), AV_OPT_TYPE_INT, {.dbl = 0}, 0, 63, VE},
  522. {NULL}
  523. };
  524. static const AVClass class = {
  525. .class_name = "libvpx encoder",
  526. .item_name = av_default_item_name,
  527. .option = options,
  528. .version = LIBAVUTIL_VERSION_INT,
  529. };
  530. static const AVCodecDefault defaults[] = {
  531. { "qmin", "-1" },
  532. { "qmax", "-1" },
  533. { "g", "-1" },
  534. { "keyint_min", "-1" },
  535. { NULL },
  536. };
  537. AVCodec ff_libvpx_encoder = {
  538. .name = "libvpx",
  539. .type = AVMEDIA_TYPE_VIDEO,
  540. .id = AV_CODEC_ID_VP8,
  541. .priv_data_size = sizeof(VP8Context),
  542. .init = vp8_init,
  543. .encode2 = vp8_encode,
  544. .close = vp8_free,
  545. .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
  546. .pix_fmts = (const enum PixelFormat[]){ PIX_FMT_YUV420P, PIX_FMT_NONE },
  547. .long_name = NULL_IF_CONFIG_SMALL("libvpx VP8"),
  548. .priv_class = &class,
  549. .defaults = defaults,
  550. };