FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
libx265.c
Go to the documentation of this file.
1 /*
2  * libx265 encoder
3  *
4  * Copyright (c) 2013-2014 Derek Buitenhuis
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg 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  * FFmpeg 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 FFmpeg; 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 #include <float.h>
29 
30 #include "libavutil/internal.h"
31 #include "libavutil/common.h"
32 #include "libavutil/opt.h"
33 #include "libavutil/pixdesc.h"
34 #include "avcodec.h"
35 #include "internal.h"
36 
37 typedef struct libx265Context {
38  const AVClass *class;
39 
40  x265_encoder *encoder;
41  x265_param *params;
42  const x265_api *api;
43 
44  float crf;
45  char *preset;
46  char *tune;
47  char *x265_opts;
49 
50 static int is_keyframe(NalUnitType naltype)
51 {
52  switch (naltype) {
53  case NAL_UNIT_CODED_SLICE_BLA_W_LP:
54  case NAL_UNIT_CODED_SLICE_BLA_W_RADL:
55  case NAL_UNIT_CODED_SLICE_BLA_N_LP:
56  case NAL_UNIT_CODED_SLICE_IDR_W_RADL:
57  case NAL_UNIT_CODED_SLICE_IDR_N_LP:
58  case NAL_UNIT_CODED_SLICE_CRA:
59  return 1;
60  default:
61  return 0;
62  }
63 }
64 
66 {
67  libx265Context *ctx = avctx->priv_data;
68 
69  av_frame_free(&avctx->coded_frame);
70 
71  ctx->api->param_free(ctx->params);
72 
73  if (ctx->encoder)
74  ctx->api->encoder_close(ctx->encoder);
75 
76  return 0;
77 }
78 
80 {
81  libx265Context *ctx = avctx->priv_data;
82 
83  ctx->api = x265_api_get(av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1);
84  if (!ctx->api)
85  ctx->api = x265_api_get(0);
86 
89  av_log(avctx, AV_LOG_ERROR,
90  "4:2:2 and 4:4:4 support is not fully defined for HEVC yet. "
91  "Set -strict experimental to encode anyway.\n");
92  return AVERROR(ENOSYS);
93  }
94 
95  avctx->coded_frame = av_frame_alloc();
96  if (!avctx->coded_frame) {
97  av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
98  return AVERROR(ENOMEM);
99  }
100 
101  ctx->params = ctx->api->param_alloc();
102  if (!ctx->params) {
103  av_log(avctx, AV_LOG_ERROR, "Could not allocate x265 param structure.\n");
104  return AVERROR(ENOMEM);
105  }
106 
107  if (ctx->api->param_default_preset(ctx->params, ctx->preset, ctx->tune) < 0) {
108  int i;
109 
110  av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", ctx->preset, ctx->tune);
111  av_log(avctx, AV_LOG_INFO, "Possible presets:");
112  for (i = 0; x265_preset_names[i]; i++)
113  av_log(avctx, AV_LOG_INFO, " %s", x265_preset_names[i]);
114 
115  av_log(avctx, AV_LOG_INFO, "\n");
116  av_log(avctx, AV_LOG_INFO, "Possible tunes:");
117  for (i = 0; x265_tune_names[i]; i++)
118  av_log(avctx, AV_LOG_INFO, " %s", x265_tune_names[i]);
119 
120  av_log(avctx, AV_LOG_INFO, "\n");
121 
122  return AVERROR(EINVAL);
123  }
124 
125  ctx->params->frameNumThreads = avctx->thread_count;
126  ctx->params->fpsNum = avctx->time_base.den;
127  ctx->params->fpsDenom = avctx->time_base.num * avctx->ticks_per_frame;
128  ctx->params->sourceWidth = avctx->width;
129  ctx->params->sourceHeight = avctx->height;
130  ctx->params->bEnablePsnr = !!(avctx->flags & CODEC_FLAG_PSNR);
131 
132  if ((avctx->color_primaries <= AVCOL_PRI_BT2020 &&
134  (avctx->color_trc <= AVCOL_TRC_BT2020_12 &&
135  avctx->color_trc != AVCOL_TRC_UNSPECIFIED) ||
136  (avctx->colorspace <= AVCOL_SPC_BT2020_CL &&
137  avctx->colorspace != AVCOL_SPC_UNSPECIFIED)) {
138 
139  ctx->params->vui.bEnableVideoSignalTypePresentFlag = 1;
140  ctx->params->vui.bEnableColorDescriptionPresentFlag = 1;
141 
142  // x265 validates the parameters internally
143  ctx->params->vui.colorPrimaries = avctx->color_primaries;
144  ctx->params->vui.transferCharacteristics = avctx->color_trc;
145  ctx->params->vui.matrixCoeffs = avctx->colorspace;
146  }
147 
148  if (avctx->sample_aspect_ratio.num > 0 && avctx->sample_aspect_ratio.den > 0) {
149  char sar[12];
150  int sar_num, sar_den;
151 
152  av_reduce(&sar_num, &sar_den,
153  avctx->sample_aspect_ratio.num,
154  avctx->sample_aspect_ratio.den, 65535);
155  snprintf(sar, sizeof(sar), "%d:%d", sar_num, sar_den);
156  if (ctx->api->param_parse(ctx->params, "sar", sar) == X265_PARAM_BAD_VALUE) {
157  av_log(avctx, AV_LOG_ERROR, "Invalid SAR: %d:%d.\n", sar_num, sar_den);
158  return AVERROR_INVALIDDATA;
159  }
160  }
161 
162  switch (avctx->pix_fmt) {
163  case AV_PIX_FMT_YUV420P:
165  ctx->params->internalCsp = X265_CSP_I420;
166  break;
167  case AV_PIX_FMT_YUV422P:
169  ctx->params->internalCsp = X265_CSP_I422;
170  break;
171  case AV_PIX_FMT_YUV444P:
173  ctx->params->internalCsp = X265_CSP_I444;
174  break;
175  }
176 
177  if (ctx->crf >= 0) {
178  char crf[6];
179 
180  snprintf(crf, sizeof(crf), "%2.2f", ctx->crf);
181  if (ctx->api->param_parse(ctx->params, "crf", crf) == X265_PARAM_BAD_VALUE) {
182  av_log(avctx, AV_LOG_ERROR, "Invalid crf: %2.2f.\n", ctx->crf);
183  return AVERROR(EINVAL);
184  }
185  } else if (avctx->bit_rate > 0) {
186  ctx->params->rc.bitrate = avctx->bit_rate / 1000;
187  ctx->params->rc.rateControlMode = X265_RC_ABR;
188  }
189 
190  if (!(avctx->flags & CODEC_FLAG_GLOBAL_HEADER))
191  ctx->params->bRepeatHeaders = 1;
192 
193  if (ctx->x265_opts) {
194  AVDictionary *dict = NULL;
195  AVDictionaryEntry *en = NULL;
196 
197  if (!av_dict_parse_string(&dict, ctx->x265_opts, "=", ":", 0)) {
198  while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
199  int parse_ret = ctx->api->param_parse(ctx->params, en->key, en->value);
200 
201  switch (parse_ret) {
202  case X265_PARAM_BAD_NAME:
203  av_log(avctx, AV_LOG_WARNING,
204  "Unknown option: %s.\n", en->key);
205  break;
206  case X265_PARAM_BAD_VALUE:
207  av_log(avctx, AV_LOG_WARNING,
208  "Invalid value for %s: %s.\n", en->key, en->value);
209  break;
210  default:
211  break;
212  }
213  }
214  av_dict_free(&dict);
215  }
216  }
217 
218  ctx->encoder = ctx->api->encoder_open(ctx->params);
219  if (!ctx->encoder) {
220  av_log(avctx, AV_LOG_ERROR, "Cannot open libx265 encoder.\n");
221  libx265_encode_close(avctx);
222  return AVERROR_INVALIDDATA;
223  }
224 
225  if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
226  x265_nal *nal;
227  int nnal;
228 
229  avctx->extradata_size = ctx->api->encoder_headers(ctx->encoder, &nal, &nnal);
230  if (avctx->extradata_size <= 0) {
231  av_log(avctx, AV_LOG_ERROR, "Cannot encode headers.\n");
232  libx265_encode_close(avctx);
233  return AVERROR_INVALIDDATA;
234  }
235 
237  if (!avctx->extradata) {
238  av_log(avctx, AV_LOG_ERROR,
239  "Cannot allocate HEVC header of size %d.\n", avctx->extradata_size);
240  libx265_encode_close(avctx);
241  return AVERROR(ENOMEM);
242  }
243 
244  memcpy(avctx->extradata, nal[0].payload, avctx->extradata_size);
245  }
246 
247  return 0;
248 }
249 
251  const AVFrame *pic, int *got_packet)
252 {
253  libx265Context *ctx = avctx->priv_data;
254  x265_picture x265pic;
255  x265_picture x265pic_out = { 0 };
256  x265_nal *nal;
257  uint8_t *dst;
258  int payload = 0;
259  int nnal;
260  int ret;
261  int i;
262 
263  ctx->api->picture_init(ctx->params, &x265pic);
264 
265  if (pic) {
266  for (i = 0; i < 3; i++) {
267  x265pic.planes[i] = pic->data[i];
268  x265pic.stride[i] = pic->linesize[i];
269  }
270 
271  x265pic.pts = pic->pts;
272  x265pic.bitDepth = av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1;
273 
274  x265pic.sliceType = pic->pict_type == AV_PICTURE_TYPE_I ? X265_TYPE_I :
275  pic->pict_type == AV_PICTURE_TYPE_P ? X265_TYPE_P :
276  pic->pict_type == AV_PICTURE_TYPE_B ? X265_TYPE_B :
277  X265_TYPE_AUTO;
278  }
279 
280  ret = ctx->api->encoder_encode(ctx->encoder, &nal, &nnal,
281  pic ? &x265pic : NULL, &x265pic_out);
282  if (ret < 0)
283  return AVERROR_EXTERNAL;
284 
285  if (!nnal)
286  return 0;
287 
288  for (i = 0; i < nnal; i++)
289  payload += nal[i].sizeBytes;
290 
291  ret = ff_alloc_packet(pkt, payload);
292  if (ret < 0) {
293  av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
294  return ret;
295  }
296  dst = pkt->data;
297 
298  for (i = 0; i < nnal; i++) {
299  memcpy(dst, nal[i].payload, nal[i].sizeBytes);
300  dst += nal[i].sizeBytes;
301 
302  if (is_keyframe(nal[i].type))
303  pkt->flags |= AV_PKT_FLAG_KEY;
304  }
305 
306  pkt->pts = x265pic_out.pts;
307  pkt->dts = x265pic_out.dts;
308 
309  switch (x265pic_out.sliceType) {
310  case X265_TYPE_IDR:
311  case X265_TYPE_I:
313  break;
314  case X265_TYPE_P:
316  break;
317  case X265_TYPE_B:
319  break;
320  }
321 
322  *got_packet = 1;
323  return 0;
324 }
325 
326 static const enum AVPixelFormat x265_csp_eight[] = {
331 };
332 
333 static const enum AVPixelFormat x265_csp_twelve[] = {
341 };
342 
344 {
345  if (x265_api_get(10))
346  codec->pix_fmts = x265_csp_twelve;
347  else if (x265_api_get(8))
348  codec->pix_fmts = x265_csp_eight;
349 }
350 
351 #define OFFSET(x) offsetof(libx265Context, x)
352 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
353 static const AVOption options[] = {
354  { "crf", "set the x265 crf", OFFSET(crf), AV_OPT_TYPE_FLOAT, { .dbl = -1 }, -1, FLT_MAX, VE },
355  { "preset", "set the x265 preset", OFFSET(preset), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
356  { "tune", "set the x265 tune parameter", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
357  { "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 },
358  { NULL }
359 };
360 
361 static const AVClass class = {
362  .class_name = "libx265",
363  .item_name = av_default_item_name,
364  .option = options,
366 };
367 
368 static const AVCodecDefault x265_defaults[] = {
369  { "b", "0" },
370  { NULL },
371 };
372 
374  .name = "libx265",
375  .long_name = NULL_IF_CONFIG_SMALL("libx265 H.265 / HEVC"),
376  .type = AVMEDIA_TYPE_VIDEO,
377  .id = AV_CODEC_ID_HEVC,
378  .init = libx265_encode_init,
379  .init_static_data = libx265_encode_init_csp,
380  .encode2 = libx265_encode_frame,
381  .close = libx265_encode_close,
382  .priv_data_size = sizeof(libx265Context),
383  .priv_class = &class,
384  .defaults = x265_defaults,
385  .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
386 };
ITU-R BT2020 for 12 bit system.
Definition: pixfmt.h:493
#define NULL
Definition: coverity.c:32
static const AVOption options[]
Definition: libx265.c:353
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2090
This structure describes decoded (raw) audio or video data.
Definition: frame.h:171
AVOption.
Definition: opt.h:255
static int is_keyframe(NalUnitType naltype)
Definition: libx265.c:50
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:68
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:182
#define LIBAVUTIL_VERSION_INT
Definition: version.h:62
float crf
Definition: libx265.c:44
AVFrame * coded_frame
the picture in the bitstream
Definition: avcodec.h:2745
int num
numerator
Definition: rational.h:44
AVRational sample_aspect_ratio
sample aspect ratio (0 if unknown) That is the width of a pixel divided by the height of the pixel...
Definition: avcodec.h:1623
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
Definition: avcodec.h:1444
x265_param * params
Definition: libx265.c:41
static AVPacket pkt
AVCodec.
Definition: avcodec.h:3181
uint8_t log2_chroma_w
Amount to shift the luma width right to find the chroma width.
Definition: pixdesc.h:80
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented...
Definition: avcodec.h:1369
#define CODEC_FLAG_PSNR
error[?] variables will be set during encoding.
Definition: avcodec.h:746
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
Definition: log.h:72
AVComponentDescriptor comp[4]
Parameters that describe how pixels are packed.
Definition: pixdesc.h:100
uint8_t
#define av_cold
Definition: attributes.h:74
#define av_malloc(s)
AVFrame * av_frame_alloc(void)
Allocate an AVFrame and set its fields to default values.
Definition: frame.c:135
AVOptions.
#define CODEC_FLAG_GLOBAL_HEADER
Place global headers in extradata instead of every keyframe.
Definition: avcodec.h:758
#define VE
Definition: libx265.c:352
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:257
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1355
AVDictionaryEntry * av_dict_get(const AVDictionary *m, const char *key, const AVDictionaryEntry *prev, int flags)
Get a dictionary entry with matching key.
Definition: dict.c:39
uint8_t * data
Definition: avcodec.h:1162
char * x265_opts
Definition: libx265.c:47
#define av_log(a,...)
#define AV_PKT_FLAG_KEY
The packet contains a keyframe.
Definition: avcodec.h:1208
uint16_t depth_minus1
Number of bits in the component minus 1.
Definition: pixdesc.h:57
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
#define CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: avcodec.h:824
av_default_item_name
#define AVERROR(e)
Definition: error.h:43
const x265_api * api
Definition: libx265.c:42
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:148
static int libx265_encode_frame(AVCodecContext *avctx, AVPacket *pkt, const AVFrame *pic, int *got_packet)
Definition: libx265.c:250
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:175
void av_dict_free(AVDictionary **pm)
Free all the memory allocated for an AVDictionary struct and all keys and values. ...
Definition: dict.c:199
int flags
CODEC_FLAG_*.
Definition: avcodec.h:1335
const char * name
Name of the codec implementation.
Definition: avcodec.h:3188
#define AV_PIX_FMT_YUV444P10
Definition: pixfmt.h:367
Libavcodec external API header.
int flags
A combination of AV_PKT_FLAG values.
Definition: avcodec.h:1168
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:67
int av_reduce(int *dst_num, int *dst_den, int64_t num, int64_t den, int64_t max)
Reduce a fraction.
Definition: rational.c:35
common internal API header
#define FF_INPUT_BUFFER_PADDING_SIZE
Required number of additionally allocated bytes at the end of the input bitstream for decoding...
Definition: avcodec.h:630
#define FF_COMPLIANCE_EXPERIMENTAL
Allow nonstandardized experimental things.
Definition: avcodec.h:2548
static av_cold void libx265_encode_init_csp(AVCodec *codec)
Definition: libx265.c:343
int bit_rate
the average bitrate
Definition: avcodec.h:1305
enum AVPixelFormat * pix_fmts
array of supported pixel formats, or NULL if unknown, array is terminated by -1
Definition: avcodec.h:3202
enum AVPictureType pict_type
Picture type of the frame.
Definition: frame.h:242
AVCodec ff_libx265_encoder
Definition: libx265.c:373
static av_cold int libx265_encode_init(AVCodecContext *avctx)
Definition: libx265.c:79
ret
Definition: avfilter.c:974
int width
picture width / height.
Definition: avcodec.h:1414
#define CODEC_CAP_AUTO_THREADS
Codec supports avctx->thread_count == 0 (auto).
Definition: avcodec.h:878
enum AVColorPrimaries color_primaries
Chromaticity coordinates of the source primaries.
Definition: avcodec.h:1939
int ff_alloc_packet(AVPacket *avpkt, int size)
Definition: utils.c:1792
int ticks_per_frame
For some codecs, the time base is closer to the field rate than the frame rate.
Definition: avcodec.h:1378
#define OFFSET(x)
Definition: libx265.c:351
int thread_count
thread count is used to decide how many independent tasks should be passed to execute() ...
Definition: avcodec.h:2753
char * preset
Definition: libx265.c:45
int av_dict_parse_string(AVDictionary **pm, const char *str, const char *key_val_sep, const char *pairs_sep, int flags)
Parse the key/value pairs list and add the parsed entries to a dictionary.
Definition: dict.c:176
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:199
main external API structure.
Definition: avcodec.h:1241
GLint GLenum type
Definition: opengl_enc.c:105
int extradata_size
Definition: avcodec.h:1356
#define AV_PIX_FMT_YUV420P10
Definition: pixfmt.h:364
Describe the class of an AVClass context structure.
Definition: log.h:67
enum AVColorSpace colorspace
YUV colorspace type.
Definition: avcodec.h:1953
enum AVColorTransferCharacteristic color_trc
Color Transfer Characteristic.
Definition: avcodec.h:1946
#define snprintf
Definition: snprintf.h:34
char * tune
Definition: libx265.c:46
#define AV_PIX_FMT_YUV422P10
Definition: pixfmt.h:365
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:182
ITU-R BT2020 constant luminance system.
Definition: pixfmt.h:511
preset
Definition: vf_curves.c:46
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:63
common internal api header.
common internal and external API header
static enum AVPixelFormat x265_csp_twelve[]
Definition: libx265.c:333
Bi-dir predicted.
Definition: avutil.h:269
char * key
Definition: dict.h:87
int den
denominator
Definition: rational.h:45
void * priv_data
Definition: avcodec.h:1283
static const AVCodecDefault x265_defaults[]
Definition: libx265.c:368
char * value
Definition: dict.h:88
static av_cold int libx265_encode_close(AVCodecContext *avctx)
Definition: libx265.c:65
static enum AVPixelFormat x265_csp_eight[]
Definition: libx265.c:326
int64_t dts
Decompression timestamp in AVStream->time_base units; the time at which the packet is decompressed...
Definition: avcodec.h:1161
ITU-R BT2020.
Definition: pixfmt.h:470
#define AV_DICT_IGNORE_SUFFIX
Return first entry in a dictionary whose first part corresponds to the search key, ignoring the suffix of the found key string.
Definition: dict.h:72
static const AVCodecDefault defaults[]
Definition: dcaenc.c:950
#define AVERROR_EXTERNAL
Generic error in an external library.
Definition: error.h:57
AVPixelFormat
Pixel format.
Definition: pixfmt.h:61
This structure stores compressed data.
Definition: avcodec.h:1139
int strict_std_compliance
strictly follow the standard (MPEG4, ...).
Definition: avcodec.h:2543
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: avcodec.h:1155
Predicted.
Definition: avutil.h:268
x265_encoder * encoder
Definition: libx265.c:40