Fix typo: CODEC_FLAG2_SSIM is in flags2, not in flags.
[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 FFmpeg.
6 *
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.
11 *
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.
16 *
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
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 x4->pic.img.i_csp = X264_CSP_I420;
93 x4->pic.img.i_plane = 3;
94
95 if (frame) {
96 for (i = 0; i < 3; i++) {
97 x4->pic.img.plane[i] = frame->data[i];
98 x4->pic.img.i_stride[i] = frame->linesize[i];
99 }
100
101 x4->pic.i_pts = frame->pts;
102 x4->pic.i_type = X264_TYPE_AUTO;
103 }
104
105 if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
106 return -1;
107
108 bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
109 if (bufsize < 0)
110 return -1;
111
112 /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
113 x4->out_pic.pts = pic_out.i_pts;
114
115 switch (pic_out.i_type) {
116 case X264_TYPE_IDR:
117 case X264_TYPE_I:
118 x4->out_pic.pict_type = FF_I_TYPE;
119 break;
120 case X264_TYPE_P:
121 x4->out_pic.pict_type = FF_P_TYPE;
122 break;
123 case X264_TYPE_B:
124 case X264_TYPE_BREF:
125 x4->out_pic.pict_type = FF_B_TYPE;
126 break;
127 }
128
129 x4->out_pic.key_frame = pic_out.b_keyframe;
130 x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
131
132 return bufsize;
133 }
134
135 static av_cold int X264_close(AVCodecContext *avctx)
136 {
137 X264Context *x4 = avctx->priv_data;
138
139 av_freep(&avctx->extradata);
140 av_free(x4->sei);
141
142 if (x4->enc)
143 x264_encoder_close(x4->enc);
144
145 return 0;
146 }
147
148 static av_cold int X264_init(AVCodecContext *avctx)
149 {
150 X264Context *x4 = avctx->priv_data;
151
152 x4->sei_size = 0;
153 x264_param_default(&x4->params);
154
155 x4->params.pf_log = X264_log;
156 x4->params.p_log_private = avctx;
157
158 x4->params.i_keyint_max = avctx->gop_size;
159 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
160 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
161 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
162 x4->params.rc.b_stat_write = avctx->flags & CODEC_FLAG_PASS1;
163 if (avctx->flags & CODEC_FLAG_PASS2) {
164 x4->params.rc.b_stat_read = 1;
165 } else {
166 if (avctx->crf) {
167 x4->params.rc.i_rc_method = X264_RC_CRF;
168 x4->params.rc.f_rf_constant = avctx->crf;
169 } else if (avctx->cqp > -1) {
170 x4->params.rc.i_rc_method = X264_RC_CQP;
171 x4->params.rc.i_qp_constant = avctx->cqp;
172 }
173 }
174
175 // if neither crf nor cqp modes are selected we have to enable the RC
176 // we do it this way because we cannot check if the bitrate has been set
177 if (!(avctx->crf || (avctx->cqp > -1)))
178 x4->params.rc.i_rc_method = X264_RC_ABR;
179
180 x4->params.i_bframe = avctx->max_b_frames;
181 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
182 x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
183 x4->params.i_bframe_bias = avctx->bframebias;
184 x4->params.i_bframe_pyramid = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
185 avctx->has_b_frames = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? 2 : !!avctx->max_b_frames;
186
187 x4->params.i_keyint_min = avctx->keyint_min;
188 if (x4->params.i_keyint_min > x4->params.i_keyint_max)
189 x4->params.i_keyint_min = x4->params.i_keyint_max;
190
191 x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
192
193 x4->params.b_deblocking_filter = avctx->flags & CODEC_FLAG_LOOP_FILTER;
194 x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
195 x4->params.i_deblocking_filter_beta = avctx->deblockbeta;
196
197 x4->params.rc.i_qp_min = avctx->qmin;
198 x4->params.rc.i_qp_max = avctx->qmax;
199 x4->params.rc.i_qp_step = avctx->max_qdiff;
200
201 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
202 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
203 x4->params.rc.f_complexity_blur = avctx->complexityblur;
204
205 x4->params.i_frame_reference = avctx->refs;
206
207 x4->params.i_width = avctx->width;
208 x4->params.i_height = avctx->height;
209 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
210 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
211 x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
212 x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
213
214 x4->params.analyse.inter = 0;
215 if (avctx->partitions) {
216 if (avctx->partitions & X264_PART_I4X4)
217 x4->params.analyse.inter |= X264_ANALYSE_I4x4;
218 if (avctx->partitions & X264_PART_I8X8)
219 x4->params.analyse.inter |= X264_ANALYSE_I8x8;
220 if (avctx->partitions & X264_PART_P8X8)
221 x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
222 if (avctx->partitions & X264_PART_P4X4)
223 x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
224 if (avctx->partitions & X264_PART_B8X8)
225 x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
226 }
227
228 x4->params.analyse.i_direct_mv_pred = avctx->directpred;
229
230 x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
231 x4->params.analyse.i_weighted_pred = avctx->weighted_p_pred;
232
233 if (avctx->me_method == ME_EPZS)
234 x4->params.analyse.i_me_method = X264_ME_DIA;
235 else if (avctx->me_method == ME_HEX)
236 x4->params.analyse.i_me_method = X264_ME_HEX;
237 else if (avctx->me_method == ME_UMH)
238 x4->params.analyse.i_me_method = X264_ME_UMH;
239 else if (avctx->me_method == ME_FULL)
240 x4->params.analyse.i_me_method = X264_ME_ESA;
241 else if (avctx->me_method == ME_TESA)
242 x4->params.analyse.i_me_method = X264_ME_TESA;
243 else x4->params.analyse.i_me_method = X264_ME_HEX;
244
245 x4->params.rc.i_aq_mode = avctx->aq_mode;
246 x4->params.rc.f_aq_strength = avctx->aq_strength;
247 x4->params.rc.i_lookahead = avctx->rc_lookahead;
248
249 x4->params.analyse.b_psy = avctx->flags2 & CODEC_FLAG2_PSY;
250 x4->params.analyse.f_psy_rd = avctx->psy_rd;
251 x4->params.analyse.f_psy_trellis = avctx->psy_trellis;
252
253 x4->params.analyse.i_me_range = avctx->me_range;
254 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
255
256 x4->params.analyse.b_mixed_references = avctx->flags2 & CODEC_FLAG2_MIXED_REFS;
257 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
258 x4->params.analyse.b_transform_8x8 = avctx->flags2 & CODEC_FLAG2_8X8DCT;
259 x4->params.analyse.b_fast_pskip = avctx->flags2 & CODEC_FLAG2_FASTPSKIP;
260
261 x4->params.analyse.i_trellis = avctx->trellis;
262 x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
263
264 if (avctx->level > 0)
265 x4->params.i_level_idc = avctx->level;
266
267 x4->params.rc.f_rate_tolerance =
268 (float)avctx->bit_rate_tolerance/avctx->bit_rate;
269
270 if ((avctx->rc_buffer_size != 0) &&
271 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
272 x4->params.rc.f_vbv_buffer_init =
273 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
274 } else
275 x4->params.rc.f_vbv_buffer_init = 0.9;
276
277 x4->params.rc.b_mb_tree = !!(avctx->flags2 & CODEC_FLAG2_MBTREE);
278 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
279 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
280 x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
281
282 x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR;
283 x4->params.analyse.b_ssim = avctx->flags2 & CODEC_FLAG2_SSIM;
284 x4->params.i_log_level = X264_LOG_DEBUG;
285
286 x4->params.b_aud = avctx->flags2 & CODEC_FLAG2_AUD;
287
288 x4->params.i_threads = avctx->thread_count;
289
290 x4->params.b_interlaced = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
291
292 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER)
293 x4->params.b_repeat_headers = 0;
294
295 x4->enc = x264_encoder_open(&x4->params);
296 if (!x4->enc)
297 return -1;
298
299 avctx->coded_frame = &x4->out_pic;
300
301 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
302 x264_nal_t *nal;
303 int nnal, s, i;
304
305 s = x264_encoder_headers(x4->enc, &nal, &nnal);
306
307 for (i = 0; i < nnal; i++)
308 if (nal[i].i_type == NAL_SEI)
309 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
310
311 avctx->extradata = av_malloc(s);
312 avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
313 }
314
315 return 0;
316 }
317
318 AVCodec libx264_encoder = {
319 .name = "libx264",
320 .type = AVMEDIA_TYPE_VIDEO,
321 .id = CODEC_ID_H264,
322 .priv_data_size = sizeof(X264Context),
323 .init = X264_init,
324 .encode = X264_frame,
325 .close = X264_close,
326 .capabilities = CODEC_CAP_DELAY,
327 .pix_fmts = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_NONE },
328 .long_name = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
329 };