FFmpeg  4.0
libopenh264enc.c
Go to the documentation of this file.
1 /*
2  * OpenH264 video encoder
3  * Copyright (C) 2014 Martin Storsjo
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 <wels/codec_api.h>
23 #include <wels/codec_ver.h>
24 
25 #include "libavutil/attributes.h"
26 #include "libavutil/common.h"
27 #include "libavutil/opt.h"
28 #include "libavutil/internal.h"
29 #include "libavutil/intreadwrite.h"
30 #include "libavutil/mathematics.h"
31 
32 #include "avcodec.h"
33 #include "internal.h"
34 #include "libopenh264.h"
35 
36 #if !OPENH264_VER_AT_LEAST(1, 6)
37 #define SM_SIZELIMITED_SLICE SM_DYN_SLICE
38 #endif
39 
40 typedef struct SVCContext {
41  const AVClass *av_class;
42  ISVCEncoder *encoder;
45  char *profile;
48  int skipped;
49  int cabac;
50 } SVCContext;
51 
52 #define OFFSET(x) offsetof(SVCContext, x)
53 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
54 static const AVOption options[] = {
55 #if OPENH264_VER_AT_LEAST(1, 6)
56  { "slice_mode", "set slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_FIXEDSLCNUM_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
57 #else
58  { "slice_mode", "set slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_AUTO_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
59 #endif
60  { "fixed", "a fixed number of slices", 0, AV_OPT_TYPE_CONST, { .i64 = SM_FIXEDSLCNUM_SLICE }, 0, 0, VE, "slice_mode" },
61 #if OPENH264_VER_AT_LEAST(1, 6)
62  { "dyn", "Size limited (compatibility name)", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
63  { "sizelimited", "Size limited", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
64 #else
65  { "rowmb", "one slice per row of macroblocks", 0, AV_OPT_TYPE_CONST, { .i64 = SM_ROWMB_SLICE }, 0, 0, VE, "slice_mode" },
66  { "auto", "automatic number of slices according to number of threads", 0, AV_OPT_TYPE_CONST, { .i64 = SM_AUTO_SLICE }, 0, 0, VE, "slice_mode" },
67  { "dyn", "Dynamic slicing", 0, AV_OPT_TYPE_CONST, { .i64 = SM_DYN_SLICE }, 0, 0, VE, "slice_mode" },
68 #endif
69  { "loopfilter", "enable loop filter", OFFSET(loopfilter), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE },
70  { "profile", "set profile restrictions", OFFSET(profile), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
71  { "max_nal_size", "set maximum NAL size in bytes", OFFSET(max_nal_size), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE },
72  { "allow_skip_frames", "allow skipping frames to hit the target bitrate", OFFSET(skip_frames), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
73  { "cabac", "Enable cabac", OFFSET(cabac), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
74  { NULL }
75 };
76 
77 static const AVClass class = {
78  .class_name = "libvo_amrwbenc",
79  .item_name = av_default_item_name,
80  .option = options,
82 };
83 
85 {
86  SVCContext *s = avctx->priv_data;
87 
88  if (s->encoder)
89  WelsDestroySVCEncoder(s->encoder);
90  if (s->skipped > 0)
91  av_log(avctx, AV_LOG_WARNING, "%d frames skipped\n", s->skipped);
92  return 0;
93 }
94 
96 {
97  SVCContext *s = avctx->priv_data;
98  SEncParamExt param = { 0 };
99  int err;
100  int log_level;
101  WelsTraceCallback callback_function;
102  AVCPBProperties *props;
103 
104  if ((err = ff_libopenh264_check_version(avctx)) < 0)
105  return err;
106 
107  if (WelsCreateSVCEncoder(&s->encoder)) {
108  av_log(avctx, AV_LOG_ERROR, "Unable to create encoder\n");
109  return AVERROR_UNKNOWN;
110  }
111 
112  // Pass all libopenh264 messages to our callback, to allow ourselves to filter them.
113  log_level = WELS_LOG_DETAIL;
114  (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_LEVEL, &log_level);
115 
116  // Set the logging callback function to one that uses av_log() (see implementation above).
117  callback_function = (WelsTraceCallback) ff_libopenh264_trace_callback;
118  (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK, &callback_function);
119 
120  // Set the AVCodecContext as the libopenh264 callback context so that it can be passed to av_log().
121  (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK_CONTEXT, &avctx);
122 
123  (*s->encoder)->GetDefaultParams(s->encoder, &param);
124 
125 #if FF_API_CODER_TYPE
127  if (!s->cabac)
128  s->cabac = avctx->coder_type == FF_CODER_TYPE_AC;
130 #endif
131 
132  param.fMaxFrameRate = 1/av_q2d(avctx->time_base);
133  param.iPicWidth = avctx->width;
134  param.iPicHeight = avctx->height;
135  param.iTargetBitrate = avctx->bit_rate;
136  param.iMaxBitrate = FFMAX(avctx->rc_max_rate, avctx->bit_rate);
137  param.iRCMode = RC_QUALITY_MODE;
138  param.iTemporalLayerNum = 1;
139  param.iSpatialLayerNum = 1;
140  param.bEnableDenoise = 0;
141  param.bEnableBackgroundDetection = 1;
142  param.bEnableAdaptiveQuant = 1;
143  param.bEnableFrameSkip = s->skip_frames;
144  param.bEnableLongTermReference = 0;
145  param.iLtrMarkPeriod = 30;
146  param.uiIntraPeriod = avctx->gop_size;
147 #if OPENH264_VER_AT_LEAST(1, 4)
148  param.eSpsPpsIdStrategy = CONSTANT_ID;
149 #else
150  param.bEnableSpsPpsIdAddition = 0;
151 #endif
152  param.bPrefixNalAddingCtrl = 0;
153  param.iLoopFilterDisableIdc = !s->loopfilter;
154  param.iEntropyCodingModeFlag = 0;
155  param.iMultipleThreadIdc = avctx->thread_count;
156  if (s->profile && !strcmp(s->profile, "main"))
157  param.iEntropyCodingModeFlag = 1;
158  else if (!s->profile && s->cabac)
159  param.iEntropyCodingModeFlag = 1;
160 
161  param.sSpatialLayers[0].iVideoWidth = param.iPicWidth;
162  param.sSpatialLayers[0].iVideoHeight = param.iPicHeight;
163  param.sSpatialLayers[0].fFrameRate = param.fMaxFrameRate;
164  param.sSpatialLayers[0].iSpatialBitrate = param.iTargetBitrate;
165  param.sSpatialLayers[0].iMaxSpatialBitrate = param.iMaxBitrate;
166 
167  if ((avctx->slices > 1) && (s->max_nal_size)) {
168  av_log(avctx, AV_LOG_ERROR,
169  "Invalid combination -slices %d and -max_nal_size %d.\n",
170  avctx->slices, s->max_nal_size);
171  return AVERROR(EINVAL);
172  }
173 
174  if (avctx->slices > 1)
175  s->slice_mode = SM_FIXEDSLCNUM_SLICE;
176 
177  if (s->max_nal_size)
179 
180 #if OPENH264_VER_AT_LEAST(1, 6)
181  param.sSpatialLayers[0].sSliceArgument.uiSliceMode = s->slice_mode;
182  param.sSpatialLayers[0].sSliceArgument.uiSliceNum = avctx->slices;
183 #else
184  param.sSpatialLayers[0].sSliceCfg.uiSliceMode = s->slice_mode;
185  param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceNum = avctx->slices;
186 #endif
187 
188  if (s->slice_mode == SM_SIZELIMITED_SLICE) {
189  if (s->max_nal_size){
190  param.uiMaxNalSize = s->max_nal_size;
191 #if OPENH264_VER_AT_LEAST(1, 6)
192  param.sSpatialLayers[0].sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
193 #else
194  param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
195 #endif
196  } else {
197  av_log(avctx, AV_LOG_ERROR, "Invalid -max_nal_size, "
198  "specify a valid max_nal_size to use -slice_mode dyn\n");
199  return AVERROR(EINVAL);
200  }
201  }
202 
203  if ((*s->encoder)->InitializeExt(s->encoder, &param) != cmResultSuccess) {
204  av_log(avctx, AV_LOG_ERROR, "Initialize failed\n");
205  return AVERROR_UNKNOWN;
206  }
207 
208  if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
209  SFrameBSInfo fbi = { 0 };
210  int i, size = 0;
211  (*s->encoder)->EncodeParameterSets(s->encoder, &fbi);
212  for (i = 0; i < fbi.sLayerInfo[0].iNalCount; i++)
213  size += fbi.sLayerInfo[0].pNalLengthInByte[i];
215  if (!avctx->extradata)
216  return AVERROR(ENOMEM);
217  avctx->extradata_size = size;
218  memcpy(avctx->extradata, fbi.sLayerInfo[0].pBsBuf, size);
219  }
220 
221  props = ff_add_cpb_side_data(avctx);
222  if (!props)
223  return AVERROR(ENOMEM);
224  props->max_bitrate = param.iMaxBitrate;
225  props->avg_bitrate = param.iTargetBitrate;
226 
227  return 0;
228 }
229 
230 static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
231  const AVFrame *frame, int *got_packet)
232 {
233  SVCContext *s = avctx->priv_data;
234  SFrameBSInfo fbi = { 0 };
235  int i, ret;
236  int encoded;
237  SSourcePicture sp = { 0 };
238  int size = 0, layer, first_layer = 0;
239  int layer_size[MAX_LAYER_NUM_OF_FRAME] = { 0 };
240 
241  sp.iColorFormat = videoFormatI420;
242  for (i = 0; i < 3; i++) {
243  sp.iStride[i] = frame->linesize[i];
244  sp.pData[i] = frame->data[i];
245  }
246  sp.iPicWidth = avctx->width;
247  sp.iPicHeight = avctx->height;
248 
249  if (frame->pict_type == AV_PICTURE_TYPE_I) {
250  (*s->encoder)->ForceIntraFrame(s->encoder, true);
251  }
252 
253  encoded = (*s->encoder)->EncodeFrame(s->encoder, &sp, &fbi);
254  if (encoded != cmResultSuccess) {
255  av_log(avctx, AV_LOG_ERROR, "EncodeFrame failed\n");
256  return AVERROR_UNKNOWN;
257  }
258  if (fbi.eFrameType == videoFrameTypeSkip) {
259  s->skipped++;
260  av_log(avctx, AV_LOG_DEBUG, "frame skipped\n");
261  return 0;
262  }
263  first_layer = 0;
264  // Normal frames are returned with one single layer, while IDR
265  // frames have two layers, where the first layer contains the SPS/PPS.
266  // If using global headers, don't include the SPS/PPS in the returned
267  // packet - thus, only return one layer.
268  if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
269  first_layer = fbi.iLayerNum - 1;
270 
271  for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
272  for (i = 0; i < fbi.sLayerInfo[layer].iNalCount; i++)
273  layer_size[layer] += fbi.sLayerInfo[layer].pNalLengthInByte[i];
274  size += layer_size[layer];
275  }
276  av_log(avctx, AV_LOG_DEBUG, "%d slices\n", fbi.sLayerInfo[fbi.iLayerNum - 1].iNalCount);
277 
278  if ((ret = ff_alloc_packet2(avctx, avpkt, size, size))) {
279  av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
280  return ret;
281  }
282  size = 0;
283  for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
284  memcpy(avpkt->data + size, fbi.sLayerInfo[layer].pBsBuf, layer_size[layer]);
285  size += layer_size[layer];
286  }
287  avpkt->pts = frame->pts;
288  if (fbi.eFrameType == videoFrameTypeIDR)
289  avpkt->flags |= AV_PKT_FLAG_KEY;
290  *got_packet = 1;
291  return 0;
292 }
293 
295  .name = "libopenh264",
296  .long_name = NULL_IF_CONFIG_SMALL("OpenH264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
297  .type = AVMEDIA_TYPE_VIDEO,
298  .id = AV_CODEC_ID_H264,
299  .priv_data_size = sizeof(SVCContext),
301  .encode2 = svc_encode_frame,
302  .close = svc_encode_close,
303  .capabilities = AV_CODEC_CAP_AUTO_THREADS,
305  .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P,
306  AV_PIX_FMT_NONE },
307  .priv_class = &class,
308  .wrapper_name = "libopenh264",
309 };
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: internal.h:48
void ff_libopenh264_trace_callback(void *ctx, int level, const char *msg)
Definition: libopenh264.c:41
#define NULL
Definition: coverity.c:32
const char * s
Definition: avisynth_c.h:768
int size
This structure describes decoded (raw) audio or video data.
Definition: frame.h:218
AVOption.
Definition: opt.h:246
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:182
int64_t bit_rate
the average bitrate
Definition: avcodec.h:1568
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
int max_bitrate
Maximum bitrate of the stream, in bits per second.
Definition: avcodec.h:1104
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:191
#define AV_CODEC_CAP_AUTO_THREADS
Codec supports avctx->thread_count == 0 (auto).
Definition: avcodec.h:1027
AVCodec.
Definition: avcodec.h:3408
Macro definitions for various function/variable attributes.
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented...
Definition: avcodec.h:1640
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
int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
Check AVPacket size and/or allocate data.
Definition: encode.c:32
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:40
#define av_cold
Definition: attributes.h:82
AVOptions.
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:311
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1618
static AVFrame * frame
uint8_t * data
Definition: avcodec.h:1430
static double av_q2d(AVRational a)
Convert an AVRational to a double.
Definition: rational.h:104
#define OFFSET(x)
#define sp
Definition: regdef.h:63
ISVCEncoder * encoder
#define av_log(a,...)
#define AV_PKT_FLAG_KEY
The packet contains a keyframe.
Definition: avcodec.h:1462
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
#define AVERROR(e)
Definition: error.h:43
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:186
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:197
const AVClass * av_class
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1598
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:236
const char * name
Name of the codec implementation.
Definition: avcodec.h:3415
static av_cold int svc_encode_close(AVCodecContext *avctx)
#define FFMAX(a, b)
Definition: common.h:94
int flags
A combination of AV_PKT_FLAG values.
Definition: avcodec.h:1436
static av_cold int svc_encode_init(AVCodecContext *avctx)
common internal API header
enum AVPictureType pict_type
Picture type of the frame.
Definition: frame.h:301
#define VE
static const AVOption options[]
int width
picture width / height.
Definition: avcodec.h:1690
attribute_deprecated int coder_type
Definition: avcodec.h:2430
int thread_count
thread count is used to decide how many independent tasks should be passed to execute() ...
Definition: avcodec.h:2769
This structure describes the bitrate properties of an encoded bitstream.
Definition: avcodec.h:1099
Libavcodec external API header.
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:249
main external API structure.
Definition: avcodec.h:1518
#define FF_CODER_TYPE_AC
Definition: avcodec.h:2423
char * profile
int extradata_size
Definition: avcodec.h:1619
Describe the class of an AVClass context structure.
Definition: log.h:67
static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet)
#define SM_SIZELIMITED_SLICE
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:266
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:232
#define AV_CODEC_FLAG_GLOBAL_HEADER
Place global headers in extradata instead of every keyframe.
Definition: avcodec.h:882
int gop_size
the number of pictures in a group of pictures, or 0 for intra_only
Definition: avcodec.h:1712
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:62
#define FF_DISABLE_DEPRECATION_WARNINGS
Definition: internal.h:84
common internal api header.
common internal and external API header
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:71
AVCPBProperties * ff_add_cpb_side_data(AVCodecContext *avctx)
Add a CPB properties side data to an encoding context.
Definition: utils.c:1936
#define AV_INPUT_BUFFER_PADDING_SIZE
Required number of additionally allocated bytes at the end of the input bitstream for decoding...
Definition: avcodec.h:773
int slices
Number of slices.
Definition: avcodec.h:2164
void * priv_data
Definition: avcodec.h:1545
#define FF_ENABLE_DEPRECATION_WARNINGS
Definition: internal.h:85
int avg_bitrate
Average bitrate of the stream, in bits per second.
Definition: avcodec.h:1114
int ff_libopenh264_check_version(void *logctx)
Definition: libopenh264.c:49
AVCodec ff_libopenh264_encoder
AVPixelFormat
Pixel format.
Definition: pixfmt.h:60
This structure stores compressed data.
Definition: avcodec.h:1407
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: avcodec.h:1423
int64_t rc_max_rate
maximum bitrate
Definition: avcodec.h:2391