2 * H.264 encoding using the x264 library
3 * Copyright (C) 2005 Mans Rullgard <mans@mansr.com>
5 * This file is part of Libav.
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.
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.
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
22 #include "libavutil/opt.h"
32 typedef struct X264Context {
51 static void X264_log(void *p, int level, const char *fmt, va_list args)
53 static const int level_map[] = {
54 [X264_LOG_ERROR] = AV_LOG_ERROR,
55 [X264_LOG_WARNING] = AV_LOG_WARNING,
56 [X264_LOG_INFO] = AV_LOG_INFO,
57 [X264_LOG_DEBUG] = AV_LOG_DEBUG
60 if (level < 0 || level > X264_LOG_DEBUG)
63 av_vlog(p, level_map[level], fmt, args);
67 static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size,
68 x264_nal_t *nals, int nnal, int skip_sei)
70 X264Context *x4 = ctx->priv_data;
74 /* Write the SEI as part of the first frame. */
75 if (x4->sei_size > 0 && nnal > 0) {
76 memcpy(p, x4->sei, x4->sei_size);
81 for (i = 0; i < nnal; i++){
82 /* Don't put the SEI in extradata. */
83 if (skip_sei && nals[i].i_type == NAL_SEI) {
84 x4->sei_size = nals[i].i_payload;
85 x4->sei = av_malloc(x4->sei_size);
86 memcpy(x4->sei, nals[i].p_payload, nals[i].i_payload);
89 memcpy(p, nals[i].p_payload, nals[i].i_payload);
90 p += nals[i].i_payload;
96 static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
97 int bufsize, void *data)
99 X264Context *x4 = ctx->priv_data;
100 AVFrame *frame = data;
103 x264_picture_t pic_out;
105 x264_picture_init( &x4->pic );
106 x4->pic.img.i_csp = X264_CSP_I420;
107 x4->pic.img.i_plane = 3;
110 for (i = 0; i < 3; i++) {
111 x4->pic.img.plane[i] = frame->data[i];
112 x4->pic.img.i_stride[i] = frame->linesize[i];
115 x4->pic.i_pts = frame->pts;
117 frame->pict_type == AV_PICTURE_TYPE_I ? X264_TYPE_KEYFRAME :
118 frame->pict_type == AV_PICTURE_TYPE_P ? X264_TYPE_P :
119 frame->pict_type == AV_PICTURE_TYPE_B ? X264_TYPE_B :
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);
128 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
131 bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
134 } while (!bufsize && !frame && x264_encoder_delayed_frames(x4->enc));
136 /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
137 x4->out_pic.pts = pic_out.i_pts;
139 switch (pic_out.i_type) {
142 x4->out_pic.pict_type = AV_PICTURE_TYPE_I;
145 x4->out_pic.pict_type = AV_PICTURE_TYPE_P;
149 x4->out_pic.pict_type = AV_PICTURE_TYPE_B;
153 x4->out_pic.key_frame = pic_out.b_keyframe;
155 x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
160 static av_cold int X264_close(AVCodecContext *avctx)
162 X264Context *x4 = avctx->priv_data;
164 av_freep(&avctx->extradata);
168 x264_encoder_close(x4->enc);
173 static av_cold int X264_init(AVCodecContext *avctx)
175 X264Context *x4 = avctx->priv_data;
178 x264_param_default(&x4->params);
180 x4->params.i_keyint_max = avctx->gop_size;
182 x4->params.i_bframe = avctx->max_b_frames;
183 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
184 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
185 x4->params.i_bframe_bias = avctx->bframebias;
186 x4->params.i_bframe_pyramid = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
187 avctx->has_b_frames = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? 2 : !!avctx->max_b_frames;
189 x4->params.i_keyint_min = avctx->keyint_min;
190 if (x4->params.i_keyint_min > x4->params.i_keyint_max)
191 x4->params.i_keyint_min = x4->params.i_keyint_max;
193 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
195 x4->params.b_deblocking_filter = avctx->flags & CODEC_FLAG_LOOP_FILTER;
196 x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
197 x4->params.i_deblocking_filter_beta = avctx->deblockbeta;
199 x4->params.rc.i_qp_min = avctx->qmin;
200 x4->params.rc.i_qp_max = avctx->qmax;
201 x4->params.rc.i_qp_step = avctx->max_qdiff;
203 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
204 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
205 x4->params.rc.f_complexity_blur = avctx->complexityblur;
207 x4->params.i_frame_reference = avctx->refs;
209 x4->params.analyse.inter = 0;
210 if (avctx->partitions) {
211 if (avctx->partitions & X264_PART_I4X4)
212 x4->params.analyse.inter |= X264_ANALYSE_I4x4;
213 if (avctx->partitions & X264_PART_I8X8)
214 x4->params.analyse.inter |= X264_ANALYSE_I8x8;
215 if (avctx->partitions & X264_PART_P8X8)
216 x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
217 if (avctx->partitions & X264_PART_P4X4)
218 x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
219 if (avctx->partitions & X264_PART_B8X8)
220 x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
223 x4->params.analyse.i_direct_mv_pred = avctx->directpred;
225 x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
226 x4->params.analyse.i_weighted_pred = avctx->weighted_p_pred;
228 if (avctx->me_method == ME_EPZS)
229 x4->params.analyse.i_me_method = X264_ME_DIA;
230 else if (avctx->me_method == ME_HEX)
231 x4->params.analyse.i_me_method = X264_ME_HEX;
232 else if (avctx->me_method == ME_UMH)
233 x4->params.analyse.i_me_method = X264_ME_UMH;
234 else if (avctx->me_method == ME_FULL)
235 x4->params.analyse.i_me_method = X264_ME_ESA;
236 else if (avctx->me_method == ME_TESA)
237 x4->params.analyse.i_me_method = X264_ME_TESA;
238 else x4->params.analyse.i_me_method = X264_ME_HEX;
240 x4->params.rc.i_lookahead = avctx->rc_lookahead;
242 x4->params.analyse.b_psy = avctx->flags2 & CODEC_FLAG2_PSY;
243 x4->params.analyse.f_psy_trellis = avctx->psy_trellis;
245 x4->params.analyse.i_me_range = avctx->me_range;
246 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
248 x4->params.analyse.b_mixed_references = avctx->flags2 & CODEC_FLAG2_MIXED_REFS;
249 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
250 x4->params.analyse.b_transform_8x8 = avctx->flags2 & CODEC_FLAG2_8X8DCT;
251 x4->params.analyse.b_fast_pskip = avctx->flags2 & CODEC_FLAG2_FASTPSKIP;
253 x4->params.analyse.i_trellis = avctx->trellis;
254 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
256 if (avctx->level > 0)
257 x4->params.i_level_idc = avctx->level;
259 if (x4->preset || x4->tune)
260 if (x264_param_default_preset(&x4->params, x4->preset, x4->tune) < 0) {
261 av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", x4->preset, x4->tune);
262 return AVERROR(EINVAL);
265 x4->params.pf_log = X264_log;
266 x4->params.p_log_private = avctx;
267 x4->params.i_log_level = X264_LOG_DEBUG;
269 x4->params.b_intra_refresh = avctx->flags2 & CODEC_FLAG2_INTRA_REFRESH;
270 if (avctx->bit_rate) {
271 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
272 x4->params.rc.i_rc_method = X264_RC_ABR;
274 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
275 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
276 x4->params.rc.b_stat_write = avctx->flags & CODEC_FLAG_PASS1;
277 if (avctx->flags & CODEC_FLAG_PASS2) {
278 x4->params.rc.b_stat_read = 1;
280 #if FF_API_X264_GLOBAL_OPTS
282 x4->params.rc.i_rc_method = X264_RC_CRF;
283 x4->params.rc.f_rf_constant = avctx->crf;
284 x4->params.rc.f_rf_constant_max = avctx->crf_max;
285 } else if (avctx->cqp > -1) {
286 x4->params.rc.i_rc_method = X264_RC_CQP;
287 x4->params.rc.i_qp_constant = avctx->cqp;
292 x4->params.rc.i_rc_method = X264_RC_CRF;
293 x4->params.rc.f_rf_constant = x4->crf;
294 } else if (x4->cqp >= 0) {
295 x4->params.rc.i_rc_method = X264_RC_CQP;
296 x4->params.rc.i_qp_constant = x4->cqp;
300 if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy &&
301 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
302 x4->params.rc.f_vbv_buffer_init =
303 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
306 x4->params.rc.b_mb_tree = !!(avctx->flags2 & CODEC_FLAG2_MBTREE);
307 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
308 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
309 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
311 #if FF_API_X264_GLOBAL_OPTS
312 if (avctx->aq_mode >= 0)
313 x4->params.rc.i_aq_mode = avctx->aq_mode;
314 if (avctx->aq_strength >= 0)
315 x4->params.rc.f_aq_strength = avctx->aq_strength;
316 if (avctx->psy_rd >= 0)
317 x4->params.analyse.f_psy_rd = avctx->psy_rd;
320 if (x4->aq_mode >= 0)
321 x4->params.rc.i_aq_mode = x4->aq_mode;
322 if (x4->aq_strength >= 0)
323 x4->params.rc.f_aq_strength = x4->aq_strength;
325 x4->params.analyse.f_psy_rd = x4->psy_rd;
328 if (x4->fastfirstpass)
329 x264_param_apply_fastfirstpass(&x4->params);
332 if (x264_param_apply_profile(&x4->params, x4->profile) < 0) {
333 av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile);
334 return AVERROR(EINVAL);
337 x4->params.i_width = avctx->width;
338 x4->params.i_height = avctx->height;
339 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
340 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
341 x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
342 x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
344 x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR;
345 x4->params.analyse.b_ssim = avctx->flags2 & CODEC_FLAG2_SSIM;
347 x4->params.b_aud = avctx->flags2 & CODEC_FLAG2_AUD;
349 x4->params.i_threads = avctx->thread_count;
351 x4->params.b_interlaced = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
353 x4->params.b_open_gop = !(avctx->flags & CODEC_FLAG_CLOSED_GOP);
355 x4->params.i_slice_count = avctx->slices;
357 x4->params.vui.b_fullrange = avctx->pix_fmt == PIX_FMT_YUVJ420P;
359 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER)
360 x4->params.b_repeat_headers = 0;
362 // update AVCodecContext with x264 parameters
363 avctx->has_b_frames = x4->params.i_bframe ?
364 x4->params.i_bframe_pyramid ? 2 : 1 : 0;
365 avctx->bit_rate = x4->params.rc.i_bitrate*1000;
366 #if FF_API_X264_GLOBAL_OPTS
367 avctx->crf = x4->params.rc.f_rf_constant;
370 x4->enc = x264_encoder_open(&x4->params);
374 avctx->coded_frame = &x4->out_pic;
376 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
380 s = x264_encoder_headers(x4->enc, &nal, &nnal);
382 for (i = 0; i < nnal; i++)
383 if (nal[i].i_type == NAL_SEI)
384 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
386 avctx->extradata = av_malloc(s);
387 avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
393 #define OFFSET(x) offsetof(X264Context, x)
394 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
395 static const AVOption options[] = {
396 { "preset", "Set the encoding preset (cf. x264 --fullhelp)", OFFSET(preset), FF_OPT_TYPE_STRING, { .str = "medium" }, 0, 0, VE},
397 { "tune", "Tune the encoding params (cf. x264 --fullhelp)", OFFSET(tune), FF_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
398 { "profile", "Set profile restrictions (cf. x264 --fullhelp) ", OFFSET(profile), FF_OPT_TYPE_STRING, { 0 }, 0, 0, VE},
399 { "fastfirstpass", "Use fast settings when encoding first pass", OFFSET(fastfirstpass), FF_OPT_TYPE_INT, { 1 }, 0, 1, VE},
400 { "crf", "Select the quality for constant quality mode", OFFSET(crf), FF_OPT_TYPE_FLOAT, {-1 }, -1, FLT_MAX, VE },
401 { "cqp", "Constant quantization parameter rate control method",OFFSET(cqp), FF_OPT_TYPE_INT, {-1 }, -1, INT_MAX, VE },
402 { "aq_mode", "AQ method", OFFSET(aq_mode), FF_OPT_TYPE_INT, {-1 }, -1, INT_MAX, VE, "aq_mode"},
403 { "none", NULL, 0, FF_OPT_TYPE_CONST, {X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" },
404 { "variance", "Variance AQ (complexity mask)", 0, FF_OPT_TYPE_CONST, {X264_AQ_VARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
405 { "autovariance", "Auto-variance AQ (experimental)", 0, FF_OPT_TYPE_CONST, {X264_AQ_AUTOVARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" },
406 { "aq_strength", "AQ strength. Reduces blocking and blurring in flat and textured areas.", OFFSET(aq_strength), FF_OPT_TYPE_FLOAT, {-1}, -1, FLT_MAX, VE},
407 { "pdy_rd", "Psy RD strength.", OFFSET(psy_rd), FF_OPT_TYPE_FLOAT, {-1 }, -1, FLT_MAX, VE},
411 static const AVClass class = {
412 .class_name = "libx264",
413 .item_name = av_default_item_name,
415 .version = LIBAVUTIL_VERSION_INT,
418 static const AVCodecDefault x264_defaults[] = {
423 AVCodec ff_libx264_encoder = {
425 .type = AVMEDIA_TYPE_VIDEO,
427 .priv_data_size = sizeof(X264Context),
429 .encode = X264_frame,
431 .capabilities = CODEC_CAP_DELAY,
432 .pix_fmts = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_YUVJ420P, PIX_FMT_NONE },
433 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
434 .priv_class = &class,
435 .defaults = x264_defaults,