libx265: Use 16-bit SAR
[libav.git] / libavcodec / libx265.c
CommitLineData
50ea9315
DB
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#include <x265.h>
24
25#include "libavutil/internal.h"
26#include "libavutil/common.h"
27#include "libavutil/opt.h"
28#include "libavutil/pixdesc.h"
29#include "avcodec.h"
30#include "internal.h"
31
8aca00cc
DB
32#if defined(_MSC_VER)
33#define X265_API_IMPORTS 1
34#endif
35
50ea9315
DB
36typedef struct libx265Context {
37 const AVClass *class;
38
39 x265_encoder *encoder;
40 x265_param *params;
50ea9315
DB
41
42 char *preset;
43 char *tune;
44 char *x265_opts;
45} libx265Context;
46
47static 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
62static av_cold int libx265_encode_close(AVCodecContext *avctx)
63{
64 libx265Context *ctx = avctx->priv_data;
65
66 av_frame_free(&avctx->coded_frame);
50ea9315
DB
67
68 x265_param_free(ctx->params);
69
70 if (ctx->encoder)
71 x265_encoder_close(ctx->encoder);
72
73 return 0;
74}
75
76static av_cold int libx265_encode_init(AVCodecContext *avctx)
77{
78 libx265Context *ctx = avctx->priv_data;
79 x265_nal *nal;
cd0ac6f6 80 char sar[12];
0f7fa48c 81 int sar_num, sar_den;
50ea9315 82 int nnal;
50ea9315 83
2142b2ef
DB
84 if (avctx->strict_std_compliance > FF_COMPLIANCE_EXPERIMENTAL &&
85 !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_w &&
86 !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_h) {
87 av_log(avctx, AV_LOG_ERROR,
88 "4:4:4 support is not fully defined for HEVC yet. "
89 "Set -strict experimental to encode anyway.\n");
90 return AVERROR(ENOSYS);
91 }
92
50ea9315
DB
93 avctx->coded_frame = av_frame_alloc();
94 if (!avctx->coded_frame) {
95 av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
96 return AVERROR(ENOMEM);
97 }
98
99 ctx->params = x265_param_alloc();
100 if (!ctx->params) {
101 av_log(avctx, AV_LOG_ERROR, "Could not allocate x265 param structure.\n");
102 return AVERROR(ENOMEM);
103 }
104
50ea9315
DB
105 if (x265_param_default_preset(ctx->params, ctx->preset, ctx->tune) < 0) {
106 av_log(avctx, AV_LOG_ERROR, "Invalid preset or tune.\n");
107 return AVERROR(EINVAL);
108 }
109
110 ctx->params->frameNumThreads = avctx->thread_count;
d00a504b
DB
111 ctx->params->fpsNum = avctx->time_base.den;
112 ctx->params->fpsDenom = avctx->time_base.num * avctx->ticks_per_frame;
50ea9315
DB
113 ctx->params->sourceWidth = avctx->width;
114 ctx->params->sourceHeight = avctx->height;
d00a504b 115
0f7fa48c
DB
116 av_reduce(&sar_num, &sar_den,
117 avctx->sample_aspect_ratio.num,
cd0ac6f6 118 avctx->sample_aspect_ratio.den, 65535);
4a222645
DB
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 }
0f7fa48c 124
d00a504b
DB
125 if (x265_max_bit_depth == 8)
126 ctx->params->internalBitDepth = 8;
127 else if (x265_max_bit_depth == 12)
128 ctx->params->internalBitDepth = 10;
50ea9315 129
2142b2ef
DB
130 switch (avctx->pix_fmt) {
131 case AV_PIX_FMT_YUV420P:
132 case AV_PIX_FMT_YUV420P10:
133 ctx->params->internalCsp = X265_CSP_I420;
134 break;
135 case AV_PIX_FMT_YUV444P:
136 case AV_PIX_FMT_YUV444P10:
137 ctx->params->internalCsp = X265_CSP_I444;
138 break;
139 }
140
50ea9315
DB
141 if (avctx->bit_rate > 0) {
142 ctx->params->rc.bitrate = avctx->bit_rate / 1000;
143 ctx->params->rc.rateControlMode = X265_RC_ABR;
144 }
145
c042cbc3
MG
146 if (!(avctx->flags & CODEC_FLAG_GLOBAL_HEADER))
147 ctx->params->bRepeatHeaders = 1;
148
50ea9315
DB
149 if (ctx->x265_opts) {
150 AVDictionary *dict = NULL;
151 AVDictionaryEntry *en = NULL;
152
153 if (!av_dict_parse_string(&dict, ctx->x265_opts, "=", ":", 0)) {
154 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
155 int parse_ret = x265_param_parse(ctx->params, en->key, en->value);
156
157 switch (parse_ret) {
158 case X265_PARAM_BAD_NAME:
159 av_log(avctx, AV_LOG_WARNING,
160 "Unknown option: %s.\n", en->key);
161 break;
162 case X265_PARAM_BAD_VALUE:
163 av_log(avctx, AV_LOG_WARNING,
164 "Invalid value for %s: %s.\n", en->key, en->value);
165 break;
166 default:
167 break;
168 }
169 }
170 av_dict_free(&dict);
171 }
172 }
173
174 ctx->encoder = x265_encoder_open(ctx->params);
175 if (!ctx->encoder) {
176 av_log(avctx, AV_LOG_ERROR, "Cannot open libx265 encoder.\n");
177 libx265_encode_close(avctx);
178 return AVERROR_INVALIDDATA;
179 }
180
60728e8b 181 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
c042cbc3
MG
182 avctx->extradata_size = x265_encoder_headers(ctx->encoder, &nal, &nnal);
183 if (avctx->extradata_size <= 0) {
184 av_log(avctx, AV_LOG_ERROR, "Cannot encode headers.\n");
185 libx265_encode_close(avctx);
186 return AVERROR_INVALIDDATA;
187 }
60728e8b 188
c042cbc3
MG
189 avctx->extradata = av_malloc(avctx->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
190 if (!avctx->extradata) {
191 av_log(avctx, AV_LOG_ERROR,
192 "Cannot allocate HEVC header of size %d.\n", avctx->extradata_size);
193 libx265_encode_close(avctx);
194 return AVERROR(ENOMEM);
195 }
196
197 memcpy(avctx->extradata, nal[0].payload, avctx->extradata_size);
60728e8b
MN
198 }
199
50ea9315
DB
200 return 0;
201}
202
203static int libx265_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
204 const AVFrame *pic, int *got_packet)
205{
206 libx265Context *ctx = avctx->priv_data;
207 x265_picture x265pic;
208 x265_picture x265pic_out = { { 0 } };
209 x265_nal *nal;
210 uint8_t *dst;
211 int payload = 0;
212 int nnal;
213 int ret;
214 int i;
215
41836c4e
DB
216 x265_picture_init(ctx->params, &x265pic);
217
50ea9315
DB
218 if (pic) {
219 for (i = 0; i < 3; i++) {
220 x265pic.planes[i] = pic->data[i];
221 x265pic.stride[i] = pic->linesize[i];
222 }
223
d00a504b
DB
224 x265pic.pts = pic->pts;
225 x265pic.bitDepth = av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1;
50ea9315
DB
226 }
227
228 ret = x265_encoder_encode(ctx->encoder, &nal, &nnal,
229 pic ? &x265pic : NULL, &x265pic_out);
230 if (ret < 0)
231 return AVERROR_UNKNOWN;
232
233 if (!nnal)
234 return 0;
235
236 for (i = 0; i < nnal; i++)
237 payload += nal[i].sizeBytes;
238
50ea9315
DB
239 ret = ff_alloc_packet(pkt, payload);
240 if (ret < 0) {
241 av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
242 return ret;
243 }
244 dst = pkt->data;
245
50ea9315
DB
246 for (i = 0; i < nnal; i++) {
247 memcpy(dst, nal[i].payload, nal[i].sizeBytes);
248 dst += nal[i].sizeBytes;
249
250 if (is_keyframe(nal[i].type))
251 pkt->flags |= AV_PKT_FLAG_KEY;
252 }
253
254 pkt->pts = x265pic_out.pts;
255 pkt->dts = x265pic_out.dts;
256
257 *got_packet = 1;
258 return 0;
259}
260
261static const enum AVPixelFormat x265_csp_eight[] = {
262 AV_PIX_FMT_YUV420P,
2142b2ef 263 AV_PIX_FMT_YUV444P,
50ea9315
DB
264 AV_PIX_FMT_NONE
265};
266
267static const enum AVPixelFormat x265_csp_twelve[] = {
268 AV_PIX_FMT_YUV420P,
2142b2ef 269 AV_PIX_FMT_YUV444P,
50ea9315 270 AV_PIX_FMT_YUV420P10,
2142b2ef 271 AV_PIX_FMT_YUV444P10,
50ea9315
DB
272 AV_PIX_FMT_NONE
273};
274
275static av_cold void libx265_encode_init_csp(AVCodec *codec)
276{
277 if (x265_max_bit_depth == 8)
278 codec->pix_fmts = x265_csp_eight;
279 else if (x265_max_bit_depth == 12)
280 codec->pix_fmts = x265_csp_twelve;
281}
282
283#define OFFSET(x) offsetof(libx265Context, x)
284#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
285static const AVOption options[] = {
286 { "preset", "set the x265 preset", OFFSET(preset), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
287 { "tune", "set the x265 tune parameter", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
288 { "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 },
289 { NULL }
290};
291
292static const AVClass class = {
293 .class_name = "libx265",
294 .item_name = av_default_item_name,
295 .option = options,
296 .version = LIBAVUTIL_VERSION_INT,
297};
298
299AVCodec ff_libx265_encoder = {
300 .name = "libx265",
301 .long_name = NULL_IF_CONFIG_SMALL("libx265 H.265 / HEVC"),
302 .type = AVMEDIA_TYPE_VIDEO,
303 .id = AV_CODEC_ID_HEVC,
304 .init = libx265_encode_init,
305 .init_static_data = libx265_encode_init_csp,
306 .encode2 = libx265_encode_frame,
307 .close = libx265_encode_close,
308 .priv_data_size = sizeof(libx265Context),
309 .priv_class = &class,
310 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
311};