e5fac0046932219b17a874634c3cf300aa258950
[libav.git] / libavcodec / libx264.c
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
22 #include "avcodec.h"
23 #include <x264.h>
24 #include <math.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28
29 typedef struct X264Context {
30 x264_param_t params;
31 x264_t *enc;
32 x264_picture_t pic;
33 uint8_t *sei;
34 int sei_size;
35 AVFrame out_pic;
36 } X264Context;
37
38 static void X264_log(void *p, int level, const char *fmt, va_list args)
39 {
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
45 };
46
47 if (level < 0 || level > X264_LOG_DEBUG)
48 return;
49
50 av_vlog(p, level_map[level], fmt, args);
51 }
52
53
54 static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size,
55 x264_nal_t *nals, int nnal, int skip_sei)
56 {
57 X264Context *x4 = ctx->priv_data;
58 uint8_t *p = buf;
59 int i;
60
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);
64 p += x4->sei_size;
65 x4->sei_size = 0;
66 }
67
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);
74 continue;
75 }
76 memcpy(p, nals[i].p_payload, nals[i].i_payload);
77 p += nals[i].i_payload;
78 }
79
80 return p - buf;
81 }
82
83 static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
84 int bufsize, void *data)
85 {
86 X264Context *x4 = ctx->priv_data;
87 AVFrame *frame = data;
88 x264_nal_t *nal;
89 int nnal, i;
90 x264_picture_t pic_out;
91
92 x264_picture_init( &x4->pic );
93 x4->pic.img.i_csp = X264_CSP_I420;
94 x4->pic.img.i_plane = 3;
95
96 if (frame) {
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];
100 }
101
102 x4->pic.i_pts = frame->pts;
103 x4->pic.i_type =
104 frame->pict_type == AV_PICTURE_TYPE_I ? X264_TYPE_KEYFRAME :
105 frame->pict_type == AV_PICTURE_TYPE_P ? X264_TYPE_P :
106 frame->pict_type == AV_PICTURE_TYPE_B ? X264_TYPE_B :
107 X264_TYPE_AUTO;
108 if (x4->params.b_tff != frame->top_field_first) {
109 x4->params.b_tff = frame->top_field_first;
110 x264_encoder_reconfig(x4->enc, &x4->params);
111 }
112 }
113
114 do {
115 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
116 return -1;
117
118 bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
119 if (bufsize < 0)
120 return -1;
121 } while (!bufsize && !frame && x264_encoder_delayed_frames(x4->enc));
122
123 /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
124 x4->out_pic.pts = pic_out.i_pts;
125
126 switch (pic_out.i_type) {
127 case X264_TYPE_IDR:
128 case X264_TYPE_I:
129 x4->out_pic.pict_type = AV_PICTURE_TYPE_I;
130 break;
131 case X264_TYPE_P:
132 x4->out_pic.pict_type = AV_PICTURE_TYPE_P;
133 break;
134 case X264_TYPE_B:
135 case X264_TYPE_BREF:
136 x4->out_pic.pict_type = AV_PICTURE_TYPE_B;
137 break;
138 }
139
140 x4->out_pic.key_frame = pic_out.b_keyframe;
141 x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
142
143 return bufsize;
144 }
145
146 static av_cold int X264_close(AVCodecContext *avctx)
147 {
148 X264Context *x4 = avctx->priv_data;
149
150 av_freep(&avctx->extradata);
151 av_free(x4->sei);
152
153 if (x4->enc)
154 x264_encoder_close(x4->enc);
155
156 return 0;
157 }
158
159 static av_cold int X264_init(AVCodecContext *avctx)
160 {
161 X264Context *x4 = avctx->priv_data;
162
163 x4->sei_size = 0;
164 x264_param_default(&x4->params);
165
166 x4->params.pf_log = X264_log;
167 x4->params.p_log_private = avctx;
168
169 x4->params.i_keyint_max = avctx->gop_size;
170 x4->params.b_intra_refresh = avctx->flags2 & CODEC_FLAG2_INTRA_REFRESH;
171 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
172 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
173 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
174 x4->params.rc.b_stat_write = avctx->flags & CODEC_FLAG_PASS1;
175 if (avctx->flags & CODEC_FLAG_PASS2) {
176 x4->params.rc.b_stat_read = 1;
177 } else {
178 if (avctx->crf) {
179 x4->params.rc.i_rc_method = X264_RC_CRF;
180 x4->params.rc.f_rf_constant = avctx->crf;
181 x4->params.rc.f_rf_constant_max = avctx->crf_max;
182 } else if (avctx->cqp > -1) {
183 x4->params.rc.i_rc_method = X264_RC_CQP;
184 x4->params.rc.i_qp_constant = avctx->cqp;
185 }
186 }
187
188 // if neither crf nor cqp modes are selected we have to enable the RC
189 // we do it this way because we cannot check if the bitrate has been set
190 if (!(avctx->crf || (avctx->cqp > -1)))
191 x4->params.rc.i_rc_method = X264_RC_ABR;
192
193 x4->params.i_bframe = avctx->max_b_frames;
194 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
195 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
196 x4->params.i_bframe_bias = avctx->bframebias;
197 x4->params.i_bframe_pyramid = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
198 avctx->has_b_frames = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? 2 : !!avctx->max_b_frames;
199
200 x4->params.i_keyint_min = avctx->keyint_min;
201 if (x4->params.i_keyint_min > x4->params.i_keyint_max)
202 x4->params.i_keyint_min = x4->params.i_keyint_max;
203
204 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
205
206 x4->params.b_deblocking_filter = avctx->flags & CODEC_FLAG_LOOP_FILTER;
207 x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
208 x4->params.i_deblocking_filter_beta = avctx->deblockbeta;
209
210 x4->params.rc.i_qp_min = avctx->qmin;
211 x4->params.rc.i_qp_max = avctx->qmax;
212 x4->params.rc.i_qp_step = avctx->max_qdiff;
213
214 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
215 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
216 x4->params.rc.f_complexity_blur = avctx->complexityblur;
217
218 x4->params.i_frame_reference = avctx->refs;
219
220 x4->params.i_width = avctx->width;
221 x4->params.i_height = avctx->height;
222 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
223 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
224 x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
225 x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
226
227 x4->params.analyse.inter = 0;
228 if (avctx->partitions) {
229 if (avctx->partitions & X264_PART_I4X4)
230 x4->params.analyse.inter |= X264_ANALYSE_I4x4;
231 if (avctx->partitions & X264_PART_I8X8)
232 x4->params.analyse.inter |= X264_ANALYSE_I8x8;
233 if (avctx->partitions & X264_PART_P8X8)
234 x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
235 if (avctx->partitions & X264_PART_P4X4)
236 x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
237 if (avctx->partitions & X264_PART_B8X8)
238 x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
239 }
240
241 x4->params.analyse.i_direct_mv_pred = avctx->directpred;
242
243 x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
244 x4->params.analyse.i_weighted_pred = avctx->weighted_p_pred;
245
246 if (avctx->me_method == ME_EPZS)
247 x4->params.analyse.i_me_method = X264_ME_DIA;
248 else if (avctx->me_method == ME_HEX)
249 x4->params.analyse.i_me_method = X264_ME_HEX;
250 else if (avctx->me_method == ME_UMH)
251 x4->params.analyse.i_me_method = X264_ME_UMH;
252 else if (avctx->me_method == ME_FULL)
253 x4->params.analyse.i_me_method = X264_ME_ESA;
254 else if (avctx->me_method == ME_TESA)
255 x4->params.analyse.i_me_method = X264_ME_TESA;
256 else x4->params.analyse.i_me_method = X264_ME_HEX;
257
258 x4->params.rc.i_aq_mode = avctx->aq_mode;
259 x4->params.rc.f_aq_strength = avctx->aq_strength;
260 x4->params.rc.i_lookahead = avctx->rc_lookahead;
261
262 x4->params.analyse.b_psy = avctx->flags2 & CODEC_FLAG2_PSY;
263 x4->params.analyse.f_psy_rd = avctx->psy_rd;
264 x4->params.analyse.f_psy_trellis = avctx->psy_trellis;
265
266 x4->params.analyse.i_me_range = avctx->me_range;
267 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
268
269 x4->params.analyse.b_mixed_references = avctx->flags2 & CODEC_FLAG2_MIXED_REFS;
270 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
271 x4->params.analyse.b_transform_8x8 = avctx->flags2 & CODEC_FLAG2_8X8DCT;
272 x4->params.analyse.b_fast_pskip = avctx->flags2 & CODEC_FLAG2_FASTPSKIP;
273
274 x4->params.analyse.i_trellis = avctx->trellis;
275 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
276
277 if (avctx->level > 0)
278 x4->params.i_level_idc = avctx->level;
279
280 if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy &&
281 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
282 x4->params.rc.f_vbv_buffer_init =
283 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
284 }
285
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;
290
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;
294
295 x4->params.b_aud = avctx->flags2 & CODEC_FLAG2_AUD;
296
297 x4->params.i_threads = avctx->thread_count;
298
299 x4->params.b_interlaced = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
300
301 x4->params.b_open_gop = !(avctx->flags & CODEC_FLAG_CLOSED_GOP);
302
303 x4->params.i_slice_count = avctx->slices;
304
305 x4->params.vui.b_fullrange = avctx->pix_fmt == PIX_FMT_YUVJ420P;
306
307 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER)
308 x4->params.b_repeat_headers = 0;
309
310 x4->enc = x264_encoder_open(&x4->params);
311 if (!x4->enc)
312 return -1;
313
314 avctx->coded_frame = &x4->out_pic;
315
316 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
317 x264_nal_t *nal;
318 int nnal, s, i;
319
320 s = x264_encoder_headers(x4->enc, &nal, &nnal);
321
322 for (i = 0; i < nnal; i++)
323 if (nal[i].i_type == NAL_SEI)
324 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
325
326 avctx->extradata = av_malloc(s);
327 avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
328 }
329
330 return 0;
331 }
332
333 AVCodec ff_libx264_encoder = {
334 .name = "libx264",
335 .type = AVMEDIA_TYPE_VIDEO,
336 .id = CODEC_ID_H264,
337 .priv_data_size = sizeof(X264Context),
338 .init = X264_init,
339 .encode = X264_frame,
340 .close = X264_close,
341 .capabilities = CODEC_CAP_DELAY,
342 .pix_fmts = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_YUVJ420P, PIX_FMT_NONE },
343 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
344 };