5f2766207925c574519ee12b025d6eb636cb26a1
[libav.git] / libavcodec / libx265.c
1 /*
2 * libx265 encoder
3 *
4 * Copyright (c) 2013-2014 Derek Buitenhuis
5 *
6 * This file is part of Libav.
7 *
8 * Libav is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * Libav is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with Libav; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 #if defined(_MSC_VER)
24 #define X265_API_IMPORTS 1
25 #endif
26
27 #include <x265.h>
28
29 #include "libavutil/internal.h"
30 #include "libavutil/common.h"
31 #include "libavutil/opt.h"
32 #include "libavutil/pixdesc.h"
33 #include "avcodec.h"
34 #include "internal.h"
35
36 typedef struct libx265Context {
37 const AVClass *class;
38
39 x265_encoder *encoder;
40 x265_param *params;
41
42 char *preset;
43 char *tune;
44 char *x265_opts;
45 } libx265Context;
46
47 static int is_keyframe(NalUnitType naltype)
48 {
49 switch (naltype) {
50 case NAL_UNIT_CODED_SLICE_BLA_W_LP:
51 case NAL_UNIT_CODED_SLICE_BLA_W_RADL:
52 case NAL_UNIT_CODED_SLICE_BLA_N_LP:
53 case NAL_UNIT_CODED_SLICE_IDR_W_RADL:
54 case NAL_UNIT_CODED_SLICE_IDR_N_LP:
55 case NAL_UNIT_CODED_SLICE_CRA:
56 return 1;
57 default:
58 return 0;
59 }
60 }
61
62 static av_cold int libx265_encode_close(AVCodecContext *avctx)
63 {
64 libx265Context *ctx = avctx->priv_data;
65
66 av_frame_free(&avctx->coded_frame);
67
68 x265_param_free(ctx->params);
69
70 if (ctx->encoder)
71 x265_encoder_close(ctx->encoder);
72
73 return 0;
74 }
75
76 static av_cold int libx265_encode_init(AVCodecContext *avctx)
77 {
78 libx265Context *ctx = avctx->priv_data;
79 x265_nal *nal;
80 char sar[12];
81 int sar_num, sar_den;
82 int nnal;
83
84 if (avctx->strict_std_compliance > FF_COMPLIANCE_EXPERIMENTAL &&
85 !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_w) {
86 av_log(avctx, AV_LOG_ERROR,
87 "4:2:2 and 4:4:4 support is not fully defined for HEVC yet. "
88 "Set -strict experimental to encode anyway.\n");
89 return AVERROR(ENOSYS);
90 }
91
92 avctx->coded_frame = av_frame_alloc();
93 if (!avctx->coded_frame) {
94 av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
95 return AVERROR(ENOMEM);
96 }
97
98 ctx->params = x265_param_alloc();
99 if (!ctx->params) {
100 av_log(avctx, AV_LOG_ERROR, "Could not allocate x265 param structure.\n");
101 return AVERROR(ENOMEM);
102 }
103
104 if (x265_param_default_preset(ctx->params, ctx->preset, ctx->tune) < 0) {
105 av_log(avctx, AV_LOG_ERROR, "Invalid preset or tune.\n");
106 return AVERROR(EINVAL);
107 }
108
109 ctx->params->frameNumThreads = avctx->thread_count;
110 ctx->params->fpsNum = avctx->time_base.den;
111 ctx->params->fpsDenom = avctx->time_base.num * avctx->ticks_per_frame;
112 ctx->params->sourceWidth = avctx->width;
113 ctx->params->sourceHeight = avctx->height;
114
115 if (avctx->sample_aspect_ratio.num > 0 && avctx->sample_aspect_ratio.den > 0) {
116 av_reduce(&sar_num, &sar_den,
117 avctx->sample_aspect_ratio.num,
118 avctx->sample_aspect_ratio.den, 65535);
119 snprintf(sar, sizeof(sar), "%d:%d", sar_num, sar_den);
120 if (x265_param_parse(ctx->params, "sar", sar) == X265_PARAM_BAD_VALUE) {
121 av_log(avctx, AV_LOG_ERROR, "Invalid SAR: %d:%d.\n", sar_num, sar_den);
122 return AVERROR_INVALIDDATA;
123 }
124 }
125
126 if (x265_max_bit_depth == 8)
127 ctx->params->internalBitDepth = 8;
128 else if (x265_max_bit_depth == 12)
129 ctx->params->internalBitDepth = 10;
130
131 switch (avctx->pix_fmt) {
132 case AV_PIX_FMT_YUV420P:
133 case AV_PIX_FMT_YUV420P10:
134 ctx->params->internalCsp = X265_CSP_I420;
135 break;
136 case AV_PIX_FMT_YUV422P:
137 case AV_PIX_FMT_YUV422P10:
138 ctx->params->internalCsp = X265_CSP_I422;
139 break;
140 case AV_PIX_FMT_YUV444P:
141 case AV_PIX_FMT_YUV444P10:
142 ctx->params->internalCsp = X265_CSP_I444;
143 break;
144 }
145
146 if (avctx->bit_rate > 0) {
147 ctx->params->rc.bitrate = avctx->bit_rate / 1000;
148 ctx->params->rc.rateControlMode = X265_RC_ABR;
149 }
150
151 if (!(avctx->flags & CODEC_FLAG_GLOBAL_HEADER))
152 ctx->params->bRepeatHeaders = 1;
153
154 if (ctx->x265_opts) {
155 AVDictionary *dict = NULL;
156 AVDictionaryEntry *en = NULL;
157
158 if (!av_dict_parse_string(&dict, ctx->x265_opts, "=", ":", 0)) {
159 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
160 int parse_ret = x265_param_parse(ctx->params, en->key, en->value);
161
162 switch (parse_ret) {
163 case X265_PARAM_BAD_NAME:
164 av_log(avctx, AV_LOG_WARNING,
165 "Unknown option: %s.\n", en->key);
166 break;
167 case X265_PARAM_BAD_VALUE:
168 av_log(avctx, AV_LOG_WARNING,
169 "Invalid value for %s: %s.\n", en->key, en->value);
170 break;
171 default:
172 break;
173 }
174 }
175 av_dict_free(&dict);
176 }
177 }
178
179 ctx->encoder = x265_encoder_open(ctx->params);
180 if (!ctx->encoder) {
181 av_log(avctx, AV_LOG_ERROR, "Cannot open libx265 encoder.\n");
182 libx265_encode_close(avctx);
183 return AVERROR_INVALIDDATA;
184 }
185
186 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
187 avctx->extradata_size = x265_encoder_headers(ctx->encoder, &nal, &nnal);
188 if (avctx->extradata_size <= 0) {
189 av_log(avctx, AV_LOG_ERROR, "Cannot encode headers.\n");
190 libx265_encode_close(avctx);
191 return AVERROR_INVALIDDATA;
192 }
193
194 avctx->extradata = av_malloc(avctx->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
195 if (!avctx->extradata) {
196 av_log(avctx, AV_LOG_ERROR,
197 "Cannot allocate HEVC header of size %d.\n", avctx->extradata_size);
198 libx265_encode_close(avctx);
199 return AVERROR(ENOMEM);
200 }
201
202 memcpy(avctx->extradata, nal[0].payload, avctx->extradata_size);
203 }
204
205 return 0;
206 }
207
208 static int libx265_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
209 const AVFrame *pic, int *got_packet)
210 {
211 libx265Context *ctx = avctx->priv_data;
212 x265_picture x265pic;
213 x265_picture x265pic_out = { { 0 } };
214 x265_nal *nal;
215 uint8_t *dst;
216 int payload = 0;
217 int nnal;
218 int ret;
219 int i;
220
221 x265_picture_init(ctx->params, &x265pic);
222
223 if (pic) {
224 for (i = 0; i < 3; i++) {
225 x265pic.planes[i] = pic->data[i];
226 x265pic.stride[i] = pic->linesize[i];
227 }
228
229 x265pic.pts = pic->pts;
230 x265pic.bitDepth = av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1;
231 }
232
233 ret = x265_encoder_encode(ctx->encoder, &nal, &nnal,
234 pic ? &x265pic : NULL, &x265pic_out);
235 if (ret < 0)
236 return AVERROR_UNKNOWN;
237
238 if (!nnal)
239 return 0;
240
241 for (i = 0; i < nnal; i++)
242 payload += nal[i].sizeBytes;
243
244 ret = ff_alloc_packet(pkt, payload);
245 if (ret < 0) {
246 av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
247 return ret;
248 }
249 dst = pkt->data;
250
251 for (i = 0; i < nnal; i++) {
252 memcpy(dst, nal[i].payload, nal[i].sizeBytes);
253 dst += nal[i].sizeBytes;
254
255 if (is_keyframe(nal[i].type))
256 pkt->flags |= AV_PKT_FLAG_KEY;
257 }
258
259 pkt->pts = x265pic_out.pts;
260 pkt->dts = x265pic_out.dts;
261
262 *got_packet = 1;
263 return 0;
264 }
265
266 static const enum AVPixelFormat x265_csp_eight[] = {
267 AV_PIX_FMT_YUV420P,
268 AV_PIX_FMT_YUV422P,
269 AV_PIX_FMT_YUV444P,
270 AV_PIX_FMT_NONE
271 };
272
273 static const enum AVPixelFormat x265_csp_twelve[] = {
274 AV_PIX_FMT_YUV420P,
275 AV_PIX_FMT_YUV422P,
276 AV_PIX_FMT_YUV444P,
277 AV_PIX_FMT_YUV420P10,
278 AV_PIX_FMT_YUV422P10,
279 AV_PIX_FMT_YUV444P10,
280 AV_PIX_FMT_NONE
281 };
282
283 static av_cold void libx265_encode_init_csp(AVCodec *codec)
284 {
285 if (x265_max_bit_depth == 8)
286 codec->pix_fmts = x265_csp_eight;
287 else if (x265_max_bit_depth == 12)
288 codec->pix_fmts = x265_csp_twelve;
289 }
290
291 #define OFFSET(x) offsetof(libx265Context, x)
292 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
293 static const AVOption options[] = {
294 { "preset", "set the x265 preset", OFFSET(preset), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
295 { "tune", "set the x265 tune parameter", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
296 { "x265-params", "set the x265 configuration using a :-separated list of key=value parameters", OFFSET(x265_opts), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
297 { NULL }
298 };
299
300 static const AVClass class = {
301 .class_name = "libx265",
302 .item_name = av_default_item_name,
303 .option = options,
304 .version = LIBAVUTIL_VERSION_INT,
305 };
306
307 AVCodec ff_libx265_encoder = {
308 .name = "libx265",
309 .long_name = NULL_IF_CONFIG_SMALL("libx265 H.265 / HEVC"),
310 .type = AVMEDIA_TYPE_VIDEO,
311 .id = AV_CODEC_ID_HEVC,
312 .init = libx265_encode_init,
313 .init_static_data = libx265_encode_init_csp,
314 .encode2 = libx265_encode_frame,
315 .close = libx265_encode_close,
316 .priv_data_size = sizeof(libx265Context),
317 .priv_class = &class,
318 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
319 };