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 /**
22 * @file
23 * VP8/9 encoder support via libvpx
24 */
25
26 #define VPX_DISABLE_CTRL_TYPECHECKS 1
27 #define VPX_CODEC_DISABLE_COMPAT 1
28 #include <vpx/vpx_encoder.h>
29 #include <vpx/vp8cx.h>
30
42
43 /**
44 * Portion of struct vpx_codec_cx_pkt from vpx_encoder.h.
45 * One encoded frame returned from the library.
46 */
48 void *
buf;
/**< compressed data buffer */
49 size_t sz;
/**< length of compressed data */
52 int64_t
pts;
/**< time stamp to show frame
53 (in timebase units) */
54 unsigned long duration;
/**< duration to show frame
55 (in timebase units) */
56 uint32_t
flags;
/**< flags for this frame */
58 int have_sse;
/**< true if we have pending sse[] */
61 };
62
63 typedef struct VPxEncoderContext {
65 struct vpx_codec_ctx encoder;
66 struct vpx_image rawimg;
67 struct vpx_codec_ctx encoder_alpha;
68 struct vpx_image rawimg_alpha;
70 struct vpx_fixed_buf twopass_stats;
73 int have_sse;
/**< true if we have pending sse[] */
76
78 /**
79 * VP8 specific flags, see VP8F_* below.
80 */
82 #define VP8F_ERROR_RESILIENT 0x00000001 ///< Enable measures appropriate for streaming over lossy links
83 #define VP8F_AUTO_ALT_REF 0x00000002
///< Enable automatic alternate reference frame generation
84
86
90
92
100
101 // VP9-only
115
116 /** String mappings for enum vp8e_enc_control_id */
118 [VP8E_SET_CPUUSED] = "VP8E_SET_CPUUSED",
119 [VP8E_SET_ENABLEAUTOALTREF] = "VP8E_SET_ENABLEAUTOALTREF",
120 [VP8E_SET_NOISE_SENSITIVITY] = "VP8E_SET_NOISE_SENSITIVITY",
121 [VP8E_SET_STATIC_THRESHOLD] = "VP8E_SET_STATIC_THRESHOLD",
122 [VP8E_SET_TOKEN_PARTITIONS] = "VP8E_SET_TOKEN_PARTITIONS",
123 [VP8E_SET_ARNR_MAXFRAMES] = "VP8E_SET_ARNR_MAXFRAMES",
124 [VP8E_SET_ARNR_STRENGTH] = "VP8E_SET_ARNR_STRENGTH",
125 [VP8E_SET_ARNR_TYPE] = "VP8E_SET_ARNR_TYPE",
126 [VP8E_SET_TUNING] = "VP8E_SET_TUNING",
127 [VP8E_SET_CQ_LEVEL] = "VP8E_SET_CQ_LEVEL",
128 [VP8E_SET_MAX_INTRA_BITRATE_PCT] = "VP8E_SET_MAX_INTRA_BITRATE_PCT",
129 #if CONFIG_LIBVPX_VP9_ENCODER
130 [VP9E_SET_LOSSLESS] = "VP9E_SET_LOSSLESS",
131 [VP9E_SET_TILE_COLUMNS] = "VP9E_SET_TILE_COLUMNS",
132 [VP9E_SET_TILE_ROWS] = "VP9E_SET_TILE_ROWS",
133 [VP9E_SET_FRAME_PARALLEL_DECODING] = "VP9E_SET_FRAME_PARALLEL_DECODING",
134 [VP9E_SET_AQ_MODE] = "VP9E_SET_AQ_MODE",
135 [VP9E_SET_COLOR_SPACE] = "VP9E_SET_COLOR_SPACE",
136 #if VPX_ENCODER_ABI_VERSION >= 11
137 [VP9E_SET_COLOR_RANGE] = "VP9E_SET_COLOR_RANGE",
138 #endif
139 #if VPX_ENCODER_ABI_VERSION >= 12
140 [VP9E_SET_TARGET_LEVEL] = "VP9E_SET_TARGET_LEVEL",
141 [VP9E_GET_LEVEL] = "VP9E_GET_LEVEL",
142 #endif
143 #ifdef VPX_CTRL_VP9E_SET_ROW_MT
144 [VP9E_SET_ROW_MT] = "VP9E_SET_ROW_MT",
145 #endif
146 #ifdef VPX_CTRL_VP9E_SET_TUNE_CONTENT
147 [VP9E_SET_TUNE_CONTENT] = "VP9E_SET_TUNE_CONTENT",
148 #endif
149 #endif
150 };
151
153 {
156 const char *detail = vpx_codec_error_detail(&ctx->
encoder);
157
159 if (detail)
161 }
162
164 const struct vpx_codec_enc_cfg *cfg)
165 {
168
169 av_log(avctx, level,
"vpx_codec_enc_cfg\n");
170 av_log(avctx, level,
"generic settings\n"
171 " %*s%u\n %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
172 #if CONFIG_LIBVPX_VP9_ENCODER
173 " %*s%u\n %*s%u\n"
174 #endif
175 " %*s{%u/%u}\n %*s%u\n %*s%d\n %*s%u\n",
176 width, "g_usage:", cfg->g_usage,
177 width, "g_threads:", cfg->g_threads,
178 width, "g_profile:", cfg->g_profile,
179 width, "g_w:", cfg->g_w,
180 width, "g_h:", cfg->g_h,
181 #if CONFIG_LIBVPX_VP9_ENCODER
182 width, "g_bit_depth:", cfg->g_bit_depth,
183 width, "g_input_bit_depth:", cfg->g_input_bit_depth,
184 #endif
185 width, "g_timebase:", cfg->g_timebase.num, cfg->g_timebase.den,
186 width, "g_error_resilient:", cfg->g_error_resilient,
187 width, "g_pass:", cfg->g_pass,
188 width, "g_lag_in_frames:", cfg->g_lag_in_frames);
189 av_log(avctx, level,
"rate control settings\n"
190 " %*s%u\n %*s%u\n %*s%u\n %*s%u\n"
192 width, "rc_dropframe_thresh:", cfg->rc_dropframe_thresh,
193 width, "rc_resize_allowed:", cfg->rc_resize_allowed,
194 width, "rc_resize_up_thresh:", cfg->rc_resize_up_thresh,
195 width, "rc_resize_down_thresh:", cfg->rc_resize_down_thresh,
196 width, "rc_end_usage:", cfg->rc_end_usage,
197 width, "rc_twopass_stats_in:", cfg->rc_twopass_stats_in.buf, cfg->rc_twopass_stats_in.sz,
198 width, "rc_target_bitrate:", cfg->rc_target_bitrate);
199 av_log(avctx, level,
"quantizer settings\n"
200 " %*s%u\n %*s%u\n",
201 width, "rc_min_quantizer:", cfg->rc_min_quantizer,
202 width, "rc_max_quantizer:", cfg->rc_max_quantizer);
203 av_log(avctx, level,
"bitrate tolerance\n"
204 " %*s%u\n %*s%u\n",
205 width, "rc_undershoot_pct:", cfg->rc_undershoot_pct,
206 width, "rc_overshoot_pct:", cfg->rc_overshoot_pct);
207 av_log(avctx, level,
"decoder buffer model\n"
208 " %*s%u\n %*s%u\n %*s%u\n",
209 width, "rc_buf_sz:", cfg->rc_buf_sz,
210 width, "rc_buf_initial_sz:", cfg->rc_buf_initial_sz,
211 width, "rc_buf_optimal_sz:", cfg->rc_buf_optimal_sz);
212 av_log(avctx, level,
"2 pass rate control settings\n"
213 " %*s%u\n %*s%u\n %*s%u\n",
214 width, "rc_2pass_vbr_bias_pct:", cfg->rc_2pass_vbr_bias_pct,
215 width, "rc_2pass_vbr_minsection_pct:", cfg->rc_2pass_vbr_minsection_pct,
216 width, "rc_2pass_vbr_maxsection_pct:", cfg->rc_2pass_vbr_maxsection_pct);
217 #if VPX_ENCODER_ABI_VERSION >= 14
218 av_log(avctx, level,
" %*s%u\n",
219 width, "rc_2pass_vbr_corpus_complexity:", cfg->rc_2pass_vbr_corpus_complexity);
220 #endif
221 av_log(avctx, level,
"keyframing settings\n"
222 " %*s%d\n %*s%u\n %*s%u\n",
223 width, "kf_mode:", cfg->kf_mode,
224 width, "kf_min_dist:", cfg->kf_min_dist,
225 width, "kf_max_dist:", cfg->kf_max_dist);
226 av_log(avctx, level,
"\n");
227 }
228
230 {
232
233 while (*p)
235 *p = cx_frame;
237 }
238
240 {
245 }
246
248 {
250
251 while (p) {
254 p = list;
255 }
256 }
257
259 enum vp8e_enc_control_id
id,
int val)
260 {
264 int res;
265
268
269 res = vpx_codec_control(&ctx->
encoder,
id, val);
270 if (res != VPX_CODEC_OK) {
271 snprintf(buf,
sizeof(buf),
"Failed to set %s codec control",
274 }
275
276 return res == VPX_CODEC_OK ? 0 :
AVERROR(EINVAL);
277 }
278
279 #if VPX_ENCODER_ABI_VERSION >= 12
281 enum vp8e_enc_control_id
id,
int *
val)
282 {
286 int res;
287
290
291 res = vpx_codec_control(&ctx->
encoder,
id, val);
292 if (res != VPX_CODEC_OK) {
293 snprintf(buf,
sizeof(buf),
"Failed to set %s codec control",
296 }
297
298 return res == VPX_CODEC_OK ? 0 :
AVERROR(EINVAL);
299 }
300 #endif
301
303 {
305
306 #if VPX_ENCODER_ABI_VERSION >= 12
309 int level_out = 0;
310 if (!codecctl_intp(avctx, VP9E_GET_LEVEL, &level_out))
312 }
313 #endif
314
315 vpx_codec_destroy(&ctx->
encoder);
321 return 0;
322 }
323
324 #if CONFIG_LIBVPX_VP9_ENCODER
326 struct vpx_codec_enc_cfg *enccfg, vpx_codec_flags_t *
flags,
327 vpx_img_fmt_t *img_fmt)
328 {
330 enccfg->g_bit_depth = enccfg->g_input_bit_depth = 8;
334 enccfg->g_profile = 0;
335 *img_fmt = VPX_IMG_FMT_I420;
336 return 0;
338 enccfg->g_profile = 1;
339 *img_fmt = VPX_IMG_FMT_I422;
340 return 0;
342 enccfg->g_profile = 1;
343 *img_fmt = VPX_IMG_FMT_I440;
344 return 0;
346 ctx->vpx_cs = VPX_CS_SRGB;
348 enccfg->g_profile = 1;
349 *img_fmt = VPX_IMG_FMT_I444;
350 return 0;
353 if (codec_caps & VPX_CODEC_CAP_HIGHBITDEPTH) {
354 enccfg->g_bit_depth = enccfg->g_input_bit_depth =
356 enccfg->g_profile = 2;
357 *img_fmt = VPX_IMG_FMT_I42016;
358 *flags |= VPX_CODEC_USE_HIGHBITDEPTH;
359 return 0;
360 }
361 break;
364 if (codec_caps & VPX_CODEC_CAP_HIGHBITDEPTH) {
365 enccfg->g_bit_depth = enccfg->g_input_bit_depth =
367 enccfg->g_profile = 3;
368 *img_fmt = VPX_IMG_FMT_I42216;
369 *flags |= VPX_CODEC_USE_HIGHBITDEPTH;
370 return 0;
371 }
372 break;
375 if (codec_caps & VPX_CODEC_CAP_HIGHBITDEPTH) {
376 enccfg->g_bit_depth = enccfg->g_input_bit_depth =
378 enccfg->g_profile = 3;
379 *img_fmt = VPX_IMG_FMT_I44016;
380 *flags |= VPX_CODEC_USE_HIGHBITDEPTH;
381 return 0;
382 }
383 break;
386 ctx->vpx_cs = VPX_CS_SRGB;
389 if (codec_caps & VPX_CODEC_CAP_HIGHBITDEPTH) {
390 enccfg->g_bit_depth = enccfg->g_input_bit_depth =
393 enccfg->g_profile = 3;
394 *img_fmt = VPX_IMG_FMT_I44416;
395 *flags |= VPX_CODEC_USE_HIGHBITDEPTH;
396 return 0;
397 }
398 break;
399 default:
400 break;
401 }
404 }
405
407 {
408 enum vpx_color_space vpx_cs;
410
413 } else {
423 default:
426 return;
427 }
428 }
430 }
431
432 #if VPX_ENCODER_ABI_VERSION >= 11
434 {
435 enum vpx_color_range vpx_cr;
440 default:
443 return;
444 }
445
447 }
448 #endif
449 #endif
450
452 const struct vpx_codec_iface *iface)
453 {
455 struct vpx_codec_enc_cfg enccfg = { 0 };
456 struct vpx_codec_enc_cfg enccfg_alpha;
459 int res;
460 vpx_img_fmt_t img_fmt = VPX_IMG_FMT_I420;
461 #if CONFIG_LIBVPX_VP9_ENCODER
462 vpx_codec_caps_t codec_caps = vpx_codec_get_caps(iface);
463 #endif
464
467
470
471 if ((res = vpx_codec_enc_config_default(iface, &enccfg, 0)) != VPX_CODEC_OK) {
473 vpx_codec_err_to_string(res));
475 }
476
477 #if CONFIG_LIBVPX_VP9_ENCODER
479 if (
set_pix_fmt(avctx, codec_caps, &enccfg, &flags, &img_fmt))
481 }
482 #endif
483
488 }
489
491
492 enccfg.g_w = avctx->
width;
493 enccfg.g_h = avctx->
height;
498
500 enccfg.g_pass = VPX_RC_FIRST_PASS;
502 enccfg.g_pass = VPX_RC_LAST_PASS;
503 else
504 enccfg.g_pass = VPX_RC_ONE_PASS;
505
508 enccfg.rc_end_usage = VPX_CBR;
509 }
else if (ctx->
crf >= 0) {
510 enccfg.rc_end_usage = VPX_CQ;
511 #if CONFIG_LIBVPX_VP9_ENCODER
513 enccfg.rc_end_usage = VPX_Q;
514 #endif
515 }
516
520 #if CONFIG_LIBVPX_VP9_ENCODER
521 } else if (enccfg.rc_end_usage == VPX_Q) {
522 #endif
523 } else {
524 if (enccfg.rc_end_usage == VPX_CQ) {
525 enccfg.rc_target_bitrate = 1000000;
526 } else {
527 avctx->
bit_rate = enccfg.rc_target_bitrate * 1000;
529 "Neither bitrate nor constrained quality specified, using default bitrate of %dkbit/sec\n",
530 enccfg.rc_target_bitrate);
531 }
532 }
533
535 enccfg.rc_min_quantizer =
536 enccfg.rc_max_quantizer = 0;
537 } else {
538 if (avctx->
qmin >= 0)
539 enccfg.rc_min_quantizer = avctx->
qmin;
540 if (avctx->
qmax >= 0)
541 enccfg.rc_max_quantizer = avctx->
qmax;
542 }
543
544 if (enccfg.rc_end_usage == VPX_CQ
545 #if CONFIG_LIBVPX_VP9_ENCODER
546 || enccfg.rc_end_usage == VPX_Q
547 #endif
548 ) {
549 if (ctx->
crf < enccfg.rc_min_quantizer || ctx->
crf > enccfg.rc_max_quantizer) {
551 "CQ level %d must be between minimum and maximum quantizer value (%d-%d)\n",
552 ctx->
crf, enccfg.rc_min_quantizer, enccfg.rc_max_quantizer);
554 }
555 }
556
557 #if FF_API_PRIVATE_OPT
562 #endif
564
565 //0-100 (0 => CBR, 100 => VBR)
568 enccfg.rc_2pass_vbr_minsection_pct =
571 enccfg.rc_2pass_vbr_maxsection_pct =
573 #if CONFIG_LIBVPX_VP9_ENCODER
575 #if VPX_ENCODER_ABI_VERSION >= 14
578 #endif
579 }
580 #endif
581
583 enccfg.rc_buf_sz =
586 enccfg.rc_buf_initial_sz =
588 enccfg.rc_buf_optimal_sz = enccfg.rc_buf_sz * 5 / 6;
593
594 //_enc_init() will balk if kf_min_dist differs from max w/VPX_KF_AUTO
598 enccfg.kf_max_dist = avctx->
gop_size;
599
600 if (enccfg.g_pass == VPX_RC_FIRST_PASS)
601 enccfg.g_lag_in_frames = 0;
602 else if (enccfg.g_pass == VPX_RC_LAST_PASS) {
603 int decode_size, ret;
604
608 }
609
612 if (ret < 0) {
617 return ret;
618 }
621 if (decode_size < 0) {
624 }
625
628 }
629
630 /* 0-3: For non-zero values the encoder increasingly optimizes for reduced
631 complexity playback on low powered devices at the expense of encode
632 quality. */
634 enccfg.g_profile = avctx->
profile;
635
637
639 /* Construct Encoder Context */
640 res = vpx_codec_enc_init(&ctx->
encoder, iface, &enccfg, flags);
641 if (res != VPX_CODEC_OK) {
644 }
645
647 enccfg_alpha = enccfg;
648 res = vpx_codec_enc_init(&ctx->
encoder_alpha, iface, &enccfg_alpha, flags);
649 if (res != VPX_CODEC_OK) {
652 }
653 }
654
655 //codec control failures are currently treated only as warnings
671
673 av_log(avctx,
AV_LOG_ERROR,
"Transparency encoding with auto_alt_ref does not work\n");
675 }
676
678 #if FF_API_PRIVATE_OPT
683 #endif
686 }
692
693 #if CONFIG_LIBVPX_VP9_ENCODER
705 set_colorspace(avctx);
706 #if VPX_ENCODER_ABI_VERSION >= 11
708 #endif
709 #if VPX_ENCODER_ABI_VERSION >= 12
711 #endif
712 #ifdef VPX_CTRL_VP9E_SET_ROW_MT
715 #endif
716 #ifdef VPX_CTRL_VP9E_SET_TUNE_CONTENT
719 #endif
720 }
721 #endif
722
724
725 //provide dummy value to initialize wrapper, values will be updated each _encode()
727 (unsigned char*)1);
728 #if CONFIG_LIBVPX_VP9_ENCODER
730 ctx->
rawimg.bit_depth = enccfg.g_bit_depth;
731 #endif
732
735 (unsigned char*)1);
736
738 if (!cpb_props)
740
741 if (enccfg.rc_end_usage == VPX_CBR ||
742 enccfg.g_pass != VPX_RC_ONE_PASS) {
746 }
748
749 return 0;
750 }
751
753 const struct vpx_codec_cx_pkt *
src,
754 const struct vpx_codec_cx_pkt *src_alpha,
756 {
757 dst->
pts = src->data.frame.pts;
758 dst->
duration = src->data.frame.duration;
759 dst->
flags = src->data.frame.flags;
760 dst->
sz = src->data.frame.sz;
761 dst->
buf = src->data.frame.buf;
763 /* For alt-ref frame, don't store PSNR or increment frame_number */
764 if (!(dst->
flags & VPX_FRAME_IS_INVISIBLE)) {
768 /* associate last-seen SSE to the frame. */
769 /* Transfers ownership from ctx to dst. */
770 /* WARNING! This makes the assumption that PSNR_PKT comes
771 just before the frame it refers to! */
772 memcpy(dst->
sse, ctx->
sse,
sizeof(dst->
sse));
774 }
775 } else {
777 }
778 if (src_alpha) {
779 dst->
buf_alpha = src_alpha->data.frame.buf;
780 dst->
sz_alpha = src_alpha->data.frame.sz;
781 } else {
784 }
785 }
786
787 /**
788 * Store coded frame information in format suitable for return from encode2().
789 *
790 * Write information from @a cx_frame to @a pkt
791 * @return packet data size on success
792 * @return a negative AVERROR on error
793 */
796 {
799 if (ret >= 0) {
800 int pict_type;
803 #if FF_API_CODED_FRAME
808 #endif
809
810 if (!!(cx_frame->
flags & VPX_FRAME_IS_KEY)) {
812 #if FF_API_CODED_FRAME
816 #endif
818 } else {
820 #if FF_API_CODED_FRAME
824 #endif
825 }
826
828 cx_frame->
have_sse ? 3 : 0, pict_type);
829
831 int i;
832 /* Beware of the Y/U/V/all order! */
833 #if FF_API_CODED_FRAME
840 #endif
841 for (i = 0; i < 3; ++i) {
842 avctx->
error[i] += cx_frame->
sse[i + 1];
843 }
845 }
850 if(!side_data) {
854 }
857 }
858 } else {
859 return ret;
860 }
862 }
863
864 /**
865 * Queue multiple output frames from the encoder, returning the front-most.
866 * In cases where vpx_codec_get_cx_data() returns more than 1 frame append
867 * the frame queue. Return the head frame if available.
868 * @return Stored frame size
869 * @return AVERROR(EINVAL) on output size error
870 * @return AVERROR(ENOMEM) on coded frame queue data allocation error
871 */
873 {
875 const struct vpx_codec_cx_pkt *
pkt;
876 const struct vpx_codec_cx_pkt *pkt_alpha =
NULL;
877 const void *iter =
NULL;
878 const void *iter_alpha =
NULL;
880
883 /* return the leading frame if we've already begun queueing */
885 if (size < 0)
889 }
890
891 /* consume all available output from the encoder before returning. buffers
892 are only good through the next vpx_codec call */
893 while ((pkt = vpx_codec_get_cx_data(&ctx->
encoder, &iter)) &&
896 switch (pkt->kind) {
897 case VPX_CODEC_CX_FRAME_PKT:
898 if (!size) {
900
901 /* avoid storing the frame when the list is empty and we haven't yet
902 provided a frame for output */
904 cx_pktcpy(&cx_frame, pkt, pkt_alpha, ctx);
906 if (size < 0)
908 } else {
911
912 if (!cx_frame) {
914 "Frame queue element alloc failed\n");
916 }
917 cx_pktcpy(cx_frame, pkt, pkt_alpha, ctx);
919
920 if (!cx_frame->
buf) {
926 }
927 memcpy(cx_frame->
buf, pkt->data.frame.buf, pkt->data.frame.sz);
936 }
937 memcpy(cx_frame->
buf_alpha, pkt_alpha->data.frame.buf, pkt_alpha->data.frame.sz);
938 }
940 }
941 break;
942 case VPX_CODEC_STATS_PKT: {
944 int err;
946 stats->sz +
947 pkt->data.twopass_stats.sz)) < 0) {
948 stats->sz = 0;
950 return err;
951 }
952 memcpy((
uint8_t*)stats->buf + stats->sz,
953 pkt->data.twopass_stats.buf, pkt->data.twopass_stats.sz);
954 stats->sz += pkt->data.twopass_stats.sz;
955 break;
956 }
957 case VPX_CODEC_PSNR_PKT:
959 ctx->
sse[0] = pkt->data.psnr.sse[0];
960 ctx->
sse[1] = pkt->data.psnr.sse[1];
961 ctx->
sse[2] = pkt->data.psnr.sse[2];
962 ctx->
sse[3] = pkt->data.psnr.sse[3];
964 break;
965 case VPX_CODEC_CUSTOM_PKT:
966 //ignore unsupported/unrecognized packet types
967 break;
968 }
969 }
970
972 }
973
976 {
978 struct vpx_image *rawimg =
NULL;
979 struct vpx_image *rawimg_alpha =
NULL;
980 int64_t timestamp = 0;
981 int res, coded_size;
982 vpx_enc_frame_flags_t flags = 0;
983
984 if (frame) {
986 rawimg->planes[VPX_PLANE_Y] = frame->
data[0];
987 rawimg->planes[VPX_PLANE_U] = frame->
data[1];
988 rawimg->planes[VPX_PLANE_V] = frame->
data[2];
989 rawimg->stride[VPX_PLANE_Y] = frame->
linesize[0];
990 rawimg->stride[VPX_PLANE_U] = frame->
linesize[1];
991 rawimg->stride[VPX_PLANE_V] = frame->
linesize[2];
995 rawimg_alpha->planes[VPX_PLANE_Y] = frame->
data[3];
998 if (!u_plane || !v_plane) {
1002 }
1004 rawimg_alpha->planes[VPX_PLANE_U] = u_plane;
1006 rawimg_alpha->planes[VPX_PLANE_V] = v_plane;
1007 rawimg_alpha->stride[VPX_PLANE_Y] = frame->
linesize[0];
1008 rawimg_alpha->stride[VPX_PLANE_U] = frame->
linesize[1];
1009 rawimg_alpha->stride[VPX_PLANE_V] = frame->
linesize[2];
1010 }
1011 timestamp = frame->
pts;
1012 #if VPX_IMAGE_ABI_VERSION >= 4
1015 rawimg->range = VPX_CR_STUDIO_RANGE;
1016 break;
1018 rawimg->range = VPX_CR_FULL_RANGE;
1019 break;
1020 }
1021 #endif
1023 flags |= VPX_EFLAG_FORCE_KF;
1024 }
1025
1026 res = vpx_codec_encode(&ctx->
encoder, rawimg, timestamp,
1028 if (res != VPX_CODEC_OK) {
1031 }
1032
1034 res = vpx_codec_encode(&ctx->
encoder_alpha, rawimg_alpha, timestamp,
1036 if (res != VPX_CODEC_OK) {
1039 }
1040 }
1041
1043
1046
1050 b64_size);
1052 }
1055 }
1056
1057 if (rawimg_alpha) {
1058 av_freep(&rawimg_alpha->planes[VPX_PLANE_U]);
1059 av_freep(&rawimg_alpha->planes[VPX_PLANE_V]);
1060 }
1061
1062 *got_packet = !!coded_size;
1063 return 0;
1064 }
1065
1066 #define OFFSET(x) offsetof(VPxContext, x)
1067 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
1068
1069 #define COMMON_OPTIONS \
1070 { "auto-alt-ref", "Enable use of alternate reference " \
1071 "frames (2-pass only)", OFFSET(auto_alt_ref), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 2, VE}, \
1072 { "lag-in-frames", "Number of frames to look ahead for " \
1073 "alternate reference frame selection", OFFSET(lag_in_frames), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE}, \
1074 { "arnr-maxframes", "altref noise reduction max frame count", OFFSET(arnr_max_frames), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE}, \
1075 { "arnr-strength", "altref noise reduction filter strength", OFFSET(arnr_strength), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE}, \
1076 { "arnr-type", "altref noise reduction filter type", OFFSET(arnr_type), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE, "arnr_type"}, \
1077 { "backward", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 1}, 0, 0, VE, "arnr_type" }, \
1078 { "forward", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 2}, 0, 0, VE, "arnr_type" }, \
1079 { "centered", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 3}, 0, 0, VE, "arnr_type" }, \
1080 { "tune", "Tune the encoding to a specific scenario", OFFSET(tune), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE, "tune"}, \
1081 { "psnr", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = VP8_TUNE_PSNR}, 0, 0, VE, "tune"}, \
1082 { "ssim", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = VP8_TUNE_SSIM}, 0, 0, VE, "tune"}, \
1083 { "deadline", "Time to spend encoding, in microseconds.", OFFSET(deadline), AV_OPT_TYPE_INT, {.i64 = VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"}, \
1084 { "best", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = VPX_DL_BEST_QUALITY}, 0, 0, VE, "quality"}, \
1085 { "good", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = VPX_DL_GOOD_QUALITY}, 0, 0, VE, "quality"}, \
1086 { "realtime", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = VPX_DL_REALTIME}, 0, 0, VE, "quality"}, \
1087 { "error-resilient", "Error resilience configuration", OFFSET(error_resilient), AV_OPT_TYPE_FLAGS, {.i64 = 0}, INT_MIN, INT_MAX, VE, "er"}, \
1088 { "max-intra-rate", "Maximum I-frame bitrate (pct) 0=unlimited", OFFSET(max_intra_rate), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VE}, \
1089 { "default", "Improve resiliency against losses of whole frames", 0, AV_OPT_TYPE_CONST, {.i64 = VPX_ERROR_RESILIENT_DEFAULT}, 0, 0, VE, "er"}, \
1090 { "partitions", "The frame partitions are independently decodable " \
1091 "by the bool decoder, meaning that partitions can be decoded even " \
1092 "though earlier partitions have been lost. Note that intra predicition" \
1093 " is still done over the partition boundary.", 0, AV_OPT_TYPE_CONST, {.i64 = VPX_ERROR_RESILIENT_PARTITIONS}, 0, 0, VE, "er"}, \
1094 { "crf", "Select the quality for constant quality mode", offsetof(VPxContext, crf), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 63, VE }, \
1095 { "static-thresh", "A change threshold on blocks below which they will be skipped by the encoder", OFFSET(static_thresh), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE }, \
1096 { "drop-threshold", "Frame drop threshold", offsetof(VPxContext, drop_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, VE }, \
1097 { "noise-sensitivity", "Noise sensitivity", OFFSET(noise_sensitivity), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 4, VE}, \
1098 { "undershoot-pct", "Datarate undershoot (min) target (%)", OFFSET(rc_undershoot_pct), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 100, VE }, \
1099 { "overshoot-pct", "Datarate overshoot (max) target (%)", OFFSET(rc_overshoot_pct), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1000, VE }, \
1100
1101 #define LEGACY_OPTIONS \
1102 {"speed", "", offsetof(VPxContext, cpu_used), AV_OPT_TYPE_INT, {.i64 = 1}, -16, 16, VE}, \
1103 {"quality", "", offsetof(VPxContext, deadline), AV_OPT_TYPE_INT, {.i64 = VPX_DL_GOOD_QUALITY}, INT_MIN, INT_MAX, VE, "quality"}, \
1104 {"vp8flags", "", offsetof(VPxContext, flags), AV_OPT_TYPE_FLAGS, {.i64 = 0}, 0, UINT_MAX, VE, "flags"}, \
1105 {"error_resilient", "enable error resilience", 0, AV_OPT_TYPE_CONST, {.i64 = VP8F_ERROR_RESILIENT}, INT_MIN, INT_MAX, VE, "flags"}, \
1106 {"altref", "enable use of alternate reference frames (VP8/2-pass only)", 0, AV_OPT_TYPE_CONST, {.i64 = VP8F_AUTO_ALT_REF}, INT_MIN, INT_MAX, VE, "flags"}, \
1107 {"arnr_max_frames", "altref noise reduction max frame count", offsetof(VPxContext, arnr_max_frames), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 15, VE}, \
1108 {"arnr_strength", "altref noise reduction filter strength", offsetof(VPxContext, arnr_strength), AV_OPT_TYPE_INT, {.i64 = 3}, 0, 6, VE}, \
1109 {"arnr_type", "altref noise reduction filter type", offsetof(VPxContext, arnr_type), AV_OPT_TYPE_INT, {.i64 = 3}, 1, 3, VE}, \
1110 {"rc_lookahead", "Number of frames to look ahead for alternate reference frame selection", offsetof(VPxContext, lag_in_frames), AV_OPT_TYPE_INT, {.i64 = 25}, 0, 25, VE}, \
1111
1112 #if CONFIG_LIBVPX_VP8_ENCODER
1113 static const AVOption vp8_options[] = {
1118 };
1119 #endif
1120
1121 #if CONFIG_LIBVPX_VP9_ENCODER
1122 static const AVOption vp9_options[] = {
1126 {
"tile-columns",
"Number of tile columns to use, log2",
OFFSET(tile_columns),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 6,
VE},
1127 {
"tile-rows",
"Number of tile rows to use, log2",
OFFSET(tile_rows),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 2,
VE},
1128 {
"frame-parallel",
"Enable frame parallel decodability features",
OFFSET(frame_parallel),
AV_OPT_TYPE_BOOL,{.i64 = -1}, -1, 1,
VE},
1129 #if VPX_ENCODER_ABI_VERSION >= 12
1130 {
"aq-mode",
"adaptive quantization mode",
OFFSET(aq_mode),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 4,
VE,
"aq_mode"},
1131 #else
1132 {
"aq-mode",
"adaptive quantization mode",
OFFSET(aq_mode),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 3,
VE,
"aq_mode"},
1133 #endif
1135 {
"variance",
"Variance based Aq", 0,
AV_OPT_TYPE_CONST, {.i64 = 1}, 0, 0,
VE,
"aq_mode" },
1136 {
"complexity",
"Complexity based Aq", 0,
AV_OPT_TYPE_CONST, {.i64 = 2}, 0, 0,
VE,
"aq_mode" },
1138 #if VPX_ENCODER_ABI_VERSION >= 12
1141 #endif
1142 #ifdef VPX_CTRL_VP9E_SET_ROW_MT
1144 #endif
1145 #ifdef VPX_CTRL_VP9E_SET_TUNE_CONTENT
1146 #if VPX_ENCODER_ABI_VERSION >= 14
1147 {
"tune-content",
"Tune content type",
OFFSET(tune_content),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 2,
VE,
"tune_content" },
1148 #else
1149 {
"tune-content",
"Tune content type",
OFFSET(tune_content),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 1,
VE,
"tune_content" },
1150 #endif
1151 {
"default",
"Regular video content", 0,
AV_OPT_TYPE_CONST, {.i64 = 0}, 0, 0,
VE,
"tune_content" },
1152 {
"screen",
"Screen capture content", 0,
AV_OPT_TYPE_CONST, {.i64 = 1}, 0, 0,
VE,
"tune_content" },
1153 #if VPX_ENCODER_ABI_VERSION >= 14
1154 {
"film",
"Film content; improves grain retention", 0,
AV_OPT_TYPE_CONST, {.i64 = 2}, 0, 0,
VE,
"tune_content" },
1155 #endif
1156 #endif
1157 #if VPX_ENCODER_ABI_VERSION >= 14
1158 {
"corpus-complexity",
"corpus vbr complexity midpoint",
OFFSET(corpus_complexity),
AV_OPT_TYPE_INT, {.i64 = -1}, -1, 10000,
VE },
1159 #endif
1162 };
1163 #endif
1164
1165 #undef COMMON_OPTIONS
1166 #undef LEGACY_OPTIONS
1167
1169 { "qmin", "-1" },
1170 { "qmax", "-1" },
1171 { "g", "-1" },
1172 { "keyint_min", "-1" },
1174 };
1175
1176 #if CONFIG_LIBVPX_VP8_ENCODER
1178 {
1179 return vpx_init(avctx, vpx_codec_vp8_cx());
1180 }
1181
1182 static const AVClass class_vp8 = {
1185 .option = vp8_options,
1187 };
1188
1200 .priv_class = &class_vp8,
1202 .wrapper_name = "libvpx",
1203 };
1204 #endif /* CONFIG_LIBVPX_VP8_ENCODER */
1205
1206 #if CONFIG_LIBVPX_VP9_ENCODER
1208 {
1209 return vpx_init(avctx, vpx_codec_vp9_cx());
1210 }
1211
1212 static const AVClass class_vp9 = {
1215 .option = vp9_options,
1217 };
1218
1220 .
name =
"libvpx-vp9",
1230 .priv_class = &class_vp9,
1233 .wrapper_name = "libvpx",
1234 };
1235 #endif /* CONFIG_LIBVPX_VP9_ENCODER */
also ITU-R BT1361 / IEC 61966-2-4 xvYCC709 / SMPTE RP177 Annex B
const char const char void * val
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
#define AV_PIX_FMT_YUV440P10
This structure describes decoded (raw) audio or video data.
static av_cold int vpx_free(AVCodecContext *avctx)
int ff_side_data_set_encoder_stats(AVPacket *pkt, int quality, int64_t *error, int error_count, int pict_type)
uint64_t error[AV_NUM_DATA_POINTERS]
error
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
#define AV_LOG_WARNING
Something somehow does not look correct.
int64_t bit_rate
the average bitrate
#define LIBAVUTIL_VERSION_INT
static av_cold int init(AVCodecContext *avctx)
struct FrameListData * coded_frame_list
int max_bitrate
Maximum bitrate of the stream, in bits per second.
also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM / IEC 61966-2-4 xvYCC601 ...
int rc_initial_buffer_occupancy
Number of bits which should be loaded into the rc buffer before decoding starts.
enum AVColorRange color_range
MPEG vs JPEG YUV range.
void * buf
compressed data buffer
#define AV_PIX_FMT_GBRP10
const char * av_default_item_name(void *ptr)
Return the context name.
also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
size_t sz
length of compressed data
static int sse(MpegEncContext *s, uint8_t *src1, uint8_t *src2, int w, int h, int stride)
#define AV_PIX_FMT_YUV420P12
char * stats_in
pass2 encoding statistics input buffer Concatenated stuff from stats_out of pass1 should be placed he...
static void cx_pktcpy(struct FrameListData *dst, const struct vpx_codec_cx_pkt *src, const struct vpx_codec_cx_pkt *src_alpha, VPxContext *ctx)
#define AV_CODEC_CAP_AUTO_THREADS
Codec supports avctx->thread_count == 0 (auto).
struct vpx_fixed_buf twopass_stats
order of coefficients is actually GBR, also IEC 61966-2-1 (sRGB)
int min_bitrate
Minimum bitrate of the stream, in bits per second.
functionally identical to above
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented...
#define VP8F_AUTO_ALT_REF
Enable automatic alternate reference frame generation.
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
#define av_assert0(cond)
assert() equivalent, that is always enabled.
int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
Check AVPacket size and/or allocate data.
planar YUV 4:2:0, 20bpp, (1 Cr & Cb sample per 2x2 Y & A samples)
struct FrameListData * next
static av_cold int codecctl_int(AVCodecContext *avctx, enum vp8e_enc_control_id id, int val)
int64_t pts
time stamp to show frame (in timebase units)
AVCodec ff_libvpx_vp8_encoder
static void coded_frame_add(void *list, struct FrameListData *cx_frame)
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
#define AV_LOG_VERBOSE
Detailed information.
int buffer_size
The size of the buffer to which the ratecontrol is applied, in bits.
#define AV_PIX_FMT_YUV422P12
char * stats_out
pass1 encoding statistics output buffer
attribute_deprecated uint64_t error[AV_NUM_DATA_POINTERS]
attribute_deprecated int frame_skip_threshold
#define AV_PKT_FLAG_KEY
The packet contains a keyframe.
static int vpx_encode(AVCodecContext *avctx, AVPacket *pkt, const AVFrame *frame, int *got_packet)
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
static int storeframe(AVCodecContext *avctx, struct FrameListData *cx_frame, AVPacket *pkt)
Store coded frame information in format suitable for return from encode2().
int qmax
maximum quantizer
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
enum AVColorRange color_range
MPEG vs JPEG YUV range.
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
int flags
AV_CODEC_FLAG_*.
Round to nearest and halfway cases away from zero.
simple assert() macros that are a bit more flexible than ISO C assert().
const char * name
Name of the codec implementation.
#define AV_PIX_FMT_YUV444P10
char * av_base64_encode(char *out, int out_size, const uint8_t *in, int in_size)
Encode data to base64 and null-terminate.
static void set_color_range(AVCodecContext *avctx)
int flags
A combination of AV_PKT_FLAG values.
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
int rc_buffer_size
decoder bitstream buffer size
int64_t rc_min_rate
minimum bitrate
common internal API header
static av_cold void dump_enc_cfg(AVCodecContext *avctx, const struct vpx_codec_enc_cfg *cfg)
enum AVPictureType pict_type
Picture type of the frame.
struct vpx_image rawimg_alpha
#define AV_BASE64_SIZE(x)
Calculate the output size needed to base64-encode x bytes to a null-terminated string.
int width
picture width / height.
#define FF_PROFILE_UNKNOWN
ITU-R BT2020 non-constant luminance system.
attribute_deprecated int noise_reduction
#define AV_CODEC_FLAG_PSNR
error[?] variables will be set during encoding.
#define AV_CODEC_FLAG_PASS1
Use internal 2pass ratecontrol in first pass mode.
static void stats(AVPacket *const *in, int n_in, unsigned *_max, unsigned *_sum)
int ticks_per_frame
For some codecs, the time base is closer to the field rate than the frame rate.
static av_cold void log_encoder_error(AVCodecContext *avctx, const char *desc)
int64_t av_rescale_rnd(int64_t a, int64_t b, int64_t c, enum AVRounding rnd)
Rescale a 64-bit integer with specified rounding.
static void error(const char *err)
int thread_count
thread count is used to decide how many independent tasks should be passed to execute() ...
the normal 2^n-1 "JPEG" YUV ranges
AVCodec ff_libvpx_vp9_encoder
#define VP8F_ERROR_RESILIENT
Enable measures appropriate for streaming over lossy links.
static av_cold int vpx_init(AVCodecContext *avctx, const struct vpx_codec_iface *iface)
This structure describes the bitrate properties of an encoded bitstream.
static const AVCodecDefault defaults[]
#define AV_LOG_INFO
Standard information.
int av_reallocp(void *ptr, size_t size)
Allocate, reallocate, or free a block of memory through a pointer to a pointer.
Libavcodec external API header.
av_cold void ff_vp9_init_static(AVCodec *codec)
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
static av_cold int vp9_init(AVFormatContext *ctx, int st_index, PayloadContext *data)
main external API structure.
static int queue_frames(AVCodecContext *avctx, AVPacket *pkt_out)
Queue multiple output frames from the encoder, returning the front-most.
void av_packet_unref(AVPacket *pkt)
Wipe the packet.
int qmin
minimum quantizer
Data found in BlockAdditional element of matroska container.
#define AV_PIX_FMT_YUV420P10
Describe the class of an AVClass context structure.
static const AVProfile profiles[]
enum AVColorSpace colorspace
YUV colorspace type.
static int set_pix_fmt(AVCodecContext *avctx, struct aom_image *img)
uint32_t flags
flags for this frame
#define AV_PIX_FMT_YUV440P12
static av_cold void free_coded_frame(struct FrameListData *cx_frame)
float qcompress
amount of qscale change between easy & hard scenes (0.0-1.0)
static enum AVPixelFormat pix_fmts[]
#define AV_PIX_FMT_GBRP12
int have_sse
true if we have pending sse[]
#define AV_PIX_FMT_YUV422P10
#define AV_PIX_FMT_YUV444P12
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
the normal 219*2^(n-8) "MPEG" YUV ranges
int flags
VP8 specific flags, see VP8F_* below.
int gop_size
the number of pictures in a group of pictures, or 0 for intra_only
static const char *const ctlidstr[]
String mappings for enum vp8e_enc_control_id.
struct vpx_codec_ctx encoder_alpha
static av_cold int vp8_init(AVFormatContext *s, int st_index, PayloadContext *vp8)
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
#define FF_DISABLE_DEPRECATION_WARNINGS
common internal api header.
common internal and external API header
struct vpx_codec_ctx encoder
attribute_deprecated AVFrame * coded_frame
the picture in the bitstream
AVCPBProperties * ff_add_cpb_side_data(AVCodecContext *avctx)
Add a CPB properties side data to an encoding context.
static av_cold void free_frame_list(struct FrameListData *list)
#define AV_CODEC_FLAG_PASS2
Use internal 2pass ratecontrol in second pass mode.
int slices
Number of slices.
#define FF_ENABLE_DEPRECATION_WARNINGS
Portion of struct vpx_codec_cx_pkt from vpx_encoder.h.
int avg_bitrate
Average bitrate of the stream, in bits per second.
int key_frame
1 -> keyframe, 0-> not
unsigned long duration
duration to show frame (in timebase units)
int av_base64_decode(uint8_t *out, const char *in_str, int out_size)
Decode a base64-encoded string.
int64_t dts
Decompression timestamp in AVStream->time_base units; the time at which the packet is decompressed...
int have_sse
true if we have pending sse[]
planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples)
uint8_t * av_packet_new_side_data(AVPacket *pkt, enum AVPacketSideDataType type, int size)
Allocate new information of a packet.
const AVProfile ff_vp9_profiles[]
AVPixelFormat
Pixel format.
This structure stores compressed data.
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
int64_t rc_max_rate
maximum bitrate
int keyint_min
minimum GOP size