2 * H.264 encoding using the x264 library
3 * Copyright (C) 2005 Mans Rullgard <mans@mansr.com>
5 * This file is part of FFmpeg.
7 * FFmpeg 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 * FFmpeg 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 FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
29 typedef struct X264Context {
38 static void X264_log(void *p, int level, const char *fmt, va_list args)
40 static const int level_map[] = {
41 [X264_LOG_ERROR] = AV_LOG_ERROR,
42 [X264_LOG_WARNING] = AV_LOG_WARNING,
43 [X264_LOG_INFO] = AV_LOG_INFO,
44 [X264_LOG_DEBUG] = AV_LOG_DEBUG
47 if (level < 0 || level > X264_LOG_DEBUG)
50 av_vlog(p, level_map[level], fmt, args);
54 static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size,
55 x264_nal_t *nals, int nnal, int skip_sei)
57 X264Context *x4 = ctx->priv_data;
61 /* Write the SEI as part of the first frame. */
62 if (x4->sei_size > 0 && nnal > 0) {
63 memcpy(p, x4->sei, x4->sei_size);
68 for (i = 0; i < nnal; i++){
69 /* Don't put the SEI in extradata. */
70 if (skip_sei && nals[i].i_type == NAL_SEI) {
71 x4->sei_size = nals[i].i_payload;
72 x4->sei = av_malloc(x4->sei_size);
73 memcpy(x4->sei, nals[i].p_payload, nals[i].i_payload);
76 memcpy(p, nals[i].p_payload, nals[i].i_payload);
77 p += nals[i].i_payload;
83 static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
84 int bufsize, void *data)
86 X264Context *x4 = ctx->priv_data;
87 AVFrame *frame = data;
90 x264_picture_t pic_out;
92 x264_picture_init( &x4->pic );
93 x4->pic.img.i_csp = X264_CSP_I420;
94 x4->pic.img.i_plane = 3;
97 for (i = 0; i < 3; i++) {
98 x4->pic.img.plane[i] = frame->data[i];
99 x4->pic.img.i_stride[i] = frame->linesize[i];
102 x4->pic.i_pts = frame->pts;
104 frame->pict_type == FF_I_TYPE ? X264_TYPE_KEYFRAME :
105 frame->pict_type == FF_P_TYPE ? X264_TYPE_P :
106 frame->pict_type == FF_B_TYPE ? X264_TYPE_B :
111 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
114 bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
117 } while (!bufsize && !frame && x264_encoder_delayed_frames(x4->enc));
119 /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
120 x4->out_pic.pts = pic_out.i_pts;
122 switch (pic_out.i_type) {
125 x4->out_pic.pict_type = FF_I_TYPE;
128 x4->out_pic.pict_type = FF_P_TYPE;
132 x4->out_pic.pict_type = FF_B_TYPE;
136 x4->out_pic.key_frame = pic_out.b_keyframe;
137 x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
142 static av_cold int X264_close(AVCodecContext *avctx)
144 X264Context *x4 = avctx->priv_data;
146 av_freep(&avctx->extradata);
150 x264_encoder_close(x4->enc);
155 static av_cold int X264_init(AVCodecContext *avctx)
157 X264Context *x4 = avctx->priv_data;
160 x264_param_default(&x4->params);
162 x4->params.pf_log = X264_log;
163 x4->params.p_log_private = avctx;
165 x4->params.i_keyint_max = avctx->gop_size;
166 x4->params.b_intra_refresh = avctx->flags2 & CODEC_FLAG2_INTRA_REFRESH;
167 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
168 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
169 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
170 x4->params.rc.b_stat_write = avctx->flags & CODEC_FLAG_PASS1;
171 if (avctx->flags & CODEC_FLAG_PASS2) {
172 x4->params.rc.b_stat_read = 1;
175 x4->params.rc.i_rc_method = X264_RC_CRF;
176 x4->params.rc.f_rf_constant = avctx->crf;
177 x4->params.rc.f_rf_constant_max = avctx->crf_max;
178 } else if (avctx->cqp > -1) {
179 x4->params.rc.i_rc_method = X264_RC_CQP;
180 x4->params.rc.i_qp_constant = avctx->cqp;
184 // if neither crf nor cqp modes are selected we have to enable the RC
185 // we do it this way because we cannot check if the bitrate has been set
186 if (!(avctx->crf || (avctx->cqp > -1)))
187 x4->params.rc.i_rc_method = X264_RC_ABR;
189 x4->params.i_bframe = avctx->max_b_frames;
190 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
191 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
192 x4->params.i_bframe_bias = avctx->bframebias;
193 x4->params.i_bframe_pyramid = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
194 avctx->has_b_frames = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? 2 : !!avctx->max_b_frames;
196 x4->params.i_keyint_min = avctx->keyint_min;
197 if (x4->params.i_keyint_min > x4->params.i_keyint_max)
198 x4->params.i_keyint_min = x4->params.i_keyint_max;
200 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
202 x4->params.b_deblocking_filter = avctx->flags & CODEC_FLAG_LOOP_FILTER;
203 x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
204 x4->params.i_deblocking_filter_beta = avctx->deblockbeta;
206 x4->params.rc.i_qp_min = avctx->qmin;
207 x4->params.rc.i_qp_max = avctx->qmax;
208 x4->params.rc.i_qp_step = avctx->max_qdiff;
210 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
211 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
212 x4->params.rc.f_complexity_blur = avctx->complexityblur;
214 x4->params.i_frame_reference = avctx->refs;
216 x4->params.i_width = avctx->width;
217 x4->params.i_height = avctx->height;
218 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
219 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
220 x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
221 x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
223 x4->params.analyse.inter = 0;
224 if (avctx->partitions) {
225 if (avctx->partitions & X264_PART_I4X4)
226 x4->params.analyse.inter |= X264_ANALYSE_I4x4;
227 if (avctx->partitions & X264_PART_I8X8)
228 x4->params.analyse.inter |= X264_ANALYSE_I8x8;
229 if (avctx->partitions & X264_PART_P8X8)
230 x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
231 if (avctx->partitions & X264_PART_P4X4)
232 x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
233 if (avctx->partitions & X264_PART_B8X8)
234 x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
237 x4->params.analyse.i_direct_mv_pred = avctx->directpred;
239 x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
240 x4->params.analyse.i_weighted_pred = avctx->weighted_p_pred;
242 if (avctx->me_method == ME_EPZS)
243 x4->params.analyse.i_me_method = X264_ME_DIA;
244 else if (avctx->me_method == ME_HEX)
245 x4->params.analyse.i_me_method = X264_ME_HEX;
246 else if (avctx->me_method == ME_UMH)
247 x4->params.analyse.i_me_method = X264_ME_UMH;
248 else if (avctx->me_method == ME_FULL)
249 x4->params.analyse.i_me_method = X264_ME_ESA;
250 else if (avctx->me_method == ME_TESA)
251 x4->params.analyse.i_me_method = X264_ME_TESA;
252 else x4->params.analyse.i_me_method = X264_ME_HEX;
254 x4->params.rc.i_aq_mode = avctx->aq_mode;
255 x4->params.rc.f_aq_strength = avctx->aq_strength;
256 x4->params.rc.i_lookahead = avctx->rc_lookahead;
258 x4->params.analyse.b_psy = avctx->flags2 & CODEC_FLAG2_PSY;
259 x4->params.analyse.f_psy_rd = avctx->psy_rd;
260 x4->params.analyse.f_psy_trellis = avctx->psy_trellis;
262 x4->params.analyse.i_me_range = avctx->me_range;
263 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
265 x4->params.analyse.b_mixed_references = avctx->flags2 & CODEC_FLAG2_MIXED_REFS;
266 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
267 x4->params.analyse.b_transform_8x8 = avctx->flags2 & CODEC_FLAG2_8X8DCT;
268 x4->params.analyse.b_fast_pskip = avctx->flags2 & CODEC_FLAG2_FASTPSKIP;
270 x4->params.analyse.i_trellis = avctx->trellis;
271 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
273 if (avctx->level > 0)
274 x4->params.i_level_idc = avctx->level;
276 x4->params.rc.f_rate_tolerance =
277 (float)avctx->bit_rate_tolerance/avctx->bit_rate;
279 if ((avctx->rc_buffer_size != 0) &&
280 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
281 x4->params.rc.f_vbv_buffer_init =
282 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
284 x4->params.rc.f_vbv_buffer_init = 0.9;
286 x4->params.rc.b_mb_tree = !!(avctx->flags2 & CODEC_FLAG2_MBTREE);
287 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
288 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
289 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
291 x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR;
292 x4->params.analyse.b_ssim = avctx->flags2 & CODEC_FLAG2_SSIM;
293 x4->params.i_log_level = X264_LOG_DEBUG;
295 x4->params.b_aud = avctx->flags2 & CODEC_FLAG2_AUD;
297 x4->params.i_threads = avctx->thread_count;
299 x4->params.b_interlaced = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
301 x4->params.i_slice_count = avctx->slices;
303 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER)
304 x4->params.b_repeat_headers = 0;
306 x4->enc = x264_encoder_open(&x4->params);
310 avctx->coded_frame = &x4->out_pic;
312 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
316 s = x264_encoder_headers(x4->enc, &nal, &nnal);
318 for (i = 0; i < nnal; i++)
319 if (nal[i].i_type == NAL_SEI)
320 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
322 avctx->extradata = av_malloc(s);
323 avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
329 AVCodec libx264_encoder = {
331 .type = AVMEDIA_TYPE_VIDEO,
333 .priv_data_size = sizeof(X264Context),
335 .encode = X264_frame,
337 .capabilities = CODEC_CAP_DELAY,
338 .pix_fmts = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_NONE },
339 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),