FFmpeg
libkvazaar.c
Go to the documentation of this file.
1 /*
2  * libkvazaar encoder
3  *
4  * Copyright (c) 2015 Tampere University of Technology
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 #include <kvazaar.h>
24 #include <stdint.h>
25 #include <string.h>
26 
27 #include "libavutil/attributes.h"
28 #include "libavutil/avassert.h"
29 #include "libavutil/dict.h"
30 #include "libavutil/error.h"
31 #include "libavutil/imgutils.h"
32 #include "libavutil/internal.h"
33 #include "libavutil/log.h"
34 #include "libavutil/mem.h"
35 #include "libavutil/pixdesc.h"
36 #include "libavutil/opt.h"
37 
38 #include "avcodec.h"
39 #include "codec_internal.h"
40 #include "encode.h"
41 #include "packet_internal.h"
42 
43 typedef struct LibkvazaarContext {
44  const AVClass *class;
45 
46  const kvz_api *api;
47  kvz_encoder *encoder;
48  kvz_config *config;
49 
50  char *kvz_params;
52 
54 {
55  LibkvazaarContext *const ctx = avctx->priv_data;
56  const kvz_api *const api = ctx->api = kvz_api_get(8);
57  kvz_config *cfg = NULL;
58  kvz_encoder *enc = NULL;
59 
60  /* Kvazaar requires width and height to be multiples of eight. */
61  if (avctx->width % 8 || avctx->height % 8) {
62  av_log(avctx, AV_LOG_ERROR,
63  "Video dimensions are not a multiple of 8 (%dx%d).\n",
64  avctx->width, avctx->height);
65  return AVERROR(ENOSYS);
66  }
67 
68  ctx->config = cfg = api->config_alloc();
69  if (!cfg) {
70  av_log(avctx, AV_LOG_ERROR,
71  "Could not allocate kvazaar config structure.\n");
72  return AVERROR(ENOMEM);
73  }
74 
75  if (!api->config_init(cfg)) {
76  av_log(avctx, AV_LOG_ERROR,
77  "Could not initialize kvazaar config structure.\n");
78  return AVERROR_BUG;
79  }
80 
81  cfg->width = avctx->width;
82  cfg->height = avctx->height;
83 
84  if (avctx->framerate.num > 0 && avctx->framerate.den > 0) {
85  cfg->framerate_num = avctx->framerate.num;
86  cfg->framerate_denom = avctx->framerate.den;
87  } else {
88  if (avctx->ticks_per_frame > INT_MAX / avctx->time_base.num) {
89  av_log(avctx, AV_LOG_ERROR,
90  "Could not set framerate for kvazaar: integer overflow\n");
91  return AVERROR(EINVAL);
92  }
93  cfg->framerate_num = avctx->time_base.den;
94  cfg->framerate_denom = avctx->time_base.num * avctx->ticks_per_frame;
95  }
96  cfg->target_bitrate = avctx->bit_rate;
97  cfg->vui.sar_width = avctx->sample_aspect_ratio.num;
98  cfg->vui.sar_height = avctx->sample_aspect_ratio.den;
99  if (avctx->bit_rate) {
100  cfg->rc_algorithm = KVZ_LAMBDA;
101  }
102 
103  if (ctx->kvz_params) {
104  AVDictionary *dict = NULL;
105  if (!av_dict_parse_string(&dict, ctx->kvz_params, "=", ",", 0)) {
106  AVDictionaryEntry *entry = NULL;
107  while ((entry = av_dict_get(dict, "", entry, AV_DICT_IGNORE_SUFFIX))) {
108  if (!api->config_parse(cfg, entry->key, entry->value)) {
109  av_log(avctx, AV_LOG_WARNING, "Invalid option: %s=%s.\n",
110  entry->key, entry->value);
111  }
112  }
113  }
114  av_dict_free(&dict);
115  }
116 
117  ctx->encoder = enc = api->encoder_open(cfg);
118  if (!enc) {
119  av_log(avctx, AV_LOG_ERROR, "Could not open kvazaar encoder.\n");
120  return AVERROR_BUG;
121  }
122 
123  if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
124  kvz_data_chunk *data_out = NULL;
125  kvz_data_chunk *chunk = NULL;
126  uint32_t len_out;
127  uint8_t *p;
128 
129  if (!api->encoder_headers(enc, &data_out, &len_out))
130  return AVERROR(ENOMEM);
131 
132  avctx->extradata = p = av_mallocz(len_out + AV_INPUT_BUFFER_PADDING_SIZE);
133  if (!p) {
134  ctx->api->chunk_free(data_out);
135  return AVERROR(ENOMEM);
136  }
137 
138  avctx->extradata_size = len_out;
139 
140  for (chunk = data_out; chunk != NULL; chunk = chunk->next) {
141  memcpy(p, chunk->data, chunk->len);
142  p += chunk->len;
143  }
144 
145  ctx->api->chunk_free(data_out);
146  }
147 
148  return 0;
149 }
150 
152 {
153  LibkvazaarContext *ctx = avctx->priv_data;
154 
155  if (ctx->api) {
156  ctx->api->encoder_close(ctx->encoder);
157  ctx->api->config_destroy(ctx->config);
158  }
159 
160  return 0;
161 }
162 
164  AVPacket *avpkt,
165  const AVFrame *frame,
166  int *got_packet_ptr)
167 {
168  LibkvazaarContext *ctx = avctx->priv_data;
169  kvz_picture *input_pic = NULL;
170  kvz_picture *recon_pic = NULL;
171  kvz_frame_info frame_info;
172  kvz_data_chunk *data_out = NULL;
173  uint32_t len_out = 0;
174  int retval = 0;
175  int pict_type;
176 
177  *got_packet_ptr = 0;
178 
179  if (frame) {
180  if (frame->width != ctx->config->width ||
181  frame->height != ctx->config->height) {
182  av_log(avctx, AV_LOG_ERROR,
183  "Changing video dimensions during encoding is not supported. "
184  "(changed from %dx%d to %dx%d)\n",
185  ctx->config->width, ctx->config->height,
186  frame->width, frame->height);
187  retval = AVERROR_INVALIDDATA;
188  goto done;
189  }
190 
191  if (frame->format != avctx->pix_fmt) {
192  av_log(avctx, AV_LOG_ERROR,
193  "Changing pixel format during encoding is not supported. "
194  "(changed from %s to %s)\n",
196  av_get_pix_fmt_name(frame->format));
197  retval = AVERROR_INVALIDDATA;
198  goto done;
199  }
200 
201  // Allocate input picture for kvazaar.
202  input_pic = ctx->api->picture_alloc(frame->width, frame->height);
203  if (!input_pic) {
204  av_log(avctx, AV_LOG_ERROR, "Failed to allocate picture.\n");
205  retval = AVERROR(ENOMEM);
206  goto done;
207  }
208 
209  // Copy pixels from frame to input_pic.
210  {
211  uint8_t *dst[4] = {
212  input_pic->data[0],
213  input_pic->data[1],
214  input_pic->data[2],
215  NULL,
216  };
217  int dst_linesizes[4] = {
218  frame->width,
219  frame->width / 2,
220  frame->width / 2,
221  0
222  };
223  av_image_copy(dst, dst_linesizes,
224  (const uint8_t **)frame->data, frame->linesize,
225  frame->format, frame->width, frame->height);
226  }
227 
228  input_pic->pts = frame->pts;
229  }
230 
231  retval = ctx->api->encoder_encode(ctx->encoder,
232  input_pic,
233  &data_out, &len_out,
234  &recon_pic, NULL,
235  &frame_info);
236  if (!retval) {
237  av_log(avctx, AV_LOG_ERROR, "Failed to encode frame.\n");
238  retval = AVERROR_INVALIDDATA;
239  goto done;
240  } else
241  retval = 0; /* kvazaar returns 1 on success */
242 
243  if (data_out) {
244  kvz_data_chunk *chunk = NULL;
245  uint64_t written = 0;
246 
247  retval = ff_get_encode_buffer(avctx, avpkt, len_out, 0);
248  if (retval < 0) {
249  av_log(avctx, AV_LOG_ERROR, "Failed to allocate output packet.\n");
250  goto done;
251  }
252 
253  for (chunk = data_out; chunk != NULL; chunk = chunk->next) {
254  av_assert0(written + chunk->len <= len_out);
255  memcpy(avpkt->data + written, chunk->data, chunk->len);
256  written += chunk->len;
257  }
258 
259  avpkt->pts = recon_pic->pts;
260  avpkt->dts = recon_pic->dts;
261  avpkt->flags = 0;
262  // IRAP VCL NAL unit types span the range
263  // [BLA_W_LP (16), RSV_IRAP_VCL23 (23)].
264  if (frame_info.nal_unit_type >= KVZ_NAL_BLA_W_LP &&
265  frame_info.nal_unit_type <= KVZ_NAL_RSV_IRAP_VCL23) {
266  avpkt->flags |= AV_PKT_FLAG_KEY;
267  }
268 
269  switch (frame_info.slice_type) {
270  case KVZ_SLICE_I:
271  pict_type = AV_PICTURE_TYPE_I;
272  break;
273  case KVZ_SLICE_P:
274  pict_type = AV_PICTURE_TYPE_P;
275  break;
276  case KVZ_SLICE_B:
277  pict_type = AV_PICTURE_TYPE_B;
278  break;
279  default:
280  av_log(avctx, AV_LOG_ERROR, "Unknown picture type encountered.\n");
281  return AVERROR_EXTERNAL;
282  }
283 
284  ff_side_data_set_encoder_stats(avpkt, frame_info.qp * FF_QP2LAMBDA, NULL, 0, pict_type);
285 
286  *got_packet_ptr = 1;
287  }
288 
289 done:
290  ctx->api->picture_free(input_pic);
291  ctx->api->picture_free(recon_pic);
292  ctx->api->chunk_free(data_out);
293  return retval;
294 }
295 
296 static const enum AVPixelFormat pix_fmts[] = {
299 };
300 
301 #define OFFSET(x) offsetof(LibkvazaarContext, x)
302 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
303 static const AVOption options[] = {
304  { "kvazaar-params", "Set kvazaar parameters as a comma-separated list of key=value pairs.",
305  OFFSET(kvz_params), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
306  { NULL },
307 };
308 
309 static const AVClass class = {
310  .class_name = "libkvazaar",
311  .item_name = av_default_item_name,
312  .option = options,
314 };
315 
316 static const FFCodecDefault defaults[] = {
317  { "b", "0" },
318  { NULL },
319 };
320 
322  .p.name = "libkvazaar",
323  CODEC_LONG_NAME("libkvazaar H.265 / HEVC"),
324  .p.type = AVMEDIA_TYPE_VIDEO,
325  .p.id = AV_CODEC_ID_HEVC,
326  .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_DELAY |
328  .p.pix_fmts = pix_fmts,
329 
330  .p.priv_class = &class,
331  .priv_data_size = sizeof(LibkvazaarContext),
332  .defaults = defaults,
333 
336  .close = libkvazaar_close,
337 
338  .caps_internal = FF_CODEC_CAP_INIT_CLEANUP |
340 
341  .p.wrapper_name = "libkvazaar",
342 };
AV_LOG_WARNING
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:186
AVPixelFormat
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
LibkvazaarContext
Definition: libkvazaar.c:43
OFFSET
#define OFFSET(x)
Definition: libkvazaar.c:301
FF_CODEC_CAP_INIT_CLEANUP
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: codec_internal.h:42
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
opt.h
ff_side_data_set_encoder_stats
int ff_side_data_set_encoder_stats(AVPacket *pkt, int quality, int64_t *error, int error_count, int pict_type)
Definition: avpacket.c:602
options
static const AVOption options[]
Definition: libkvazaar.c:303
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:330
pixdesc.h
AVPacket::data
uint8_t * data
Definition: packet.h:374
AVOption
AVOption.
Definition: opt.h:251
encode.h
LibkvazaarContext::kvz_params
char * kvz_params
Definition: libkvazaar.c:50
AV_DICT_IGNORE_SUFFIX
#define AV_DICT_IGNORE_SUFFIX
Return first entry in a dictionary whose first part corresponds to the search key,...
Definition: dict.h:75
FFCodec
Definition: codec_internal.h:127
AVDictionary
Definition: dict.c:32
AV_PKT_FLAG_KEY
#define AV_PKT_FLAG_KEY
The packet contains a keyframe.
Definition: packet.h:429
defaults
static const FFCodecDefault defaults[]
Definition: libkvazaar.c:316
AV_CODEC_FLAG_GLOBAL_HEADER
#define AV_CODEC_FLAG_GLOBAL_HEADER
Place global headers in extradata instead of every keyframe.
Definition: avcodec.h:317
AVCodecContext::framerate
AVRational framerate
Definition: avcodec.h:1750
FFCodecDefault
Definition: codec_internal.h:97
FFCodec::p
AVCodec p
The public AVCodec.
Definition: codec_internal.h:131
libkvazaar_init
static av_cold int libkvazaar_init(AVCodecContext *avctx)
Definition: libkvazaar.c:53
AVCodecContext::flags
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:506
ff_libkvazaar_encoder
const FFCodec ff_libkvazaar_encoder
Definition: libkvazaar.c:321
FF_CODEC_ENCODE_CB
#define FF_CODEC_ENCODE_CB(func)
Definition: codec_internal.h:315
AVRational::num
int num
Numerator.
Definition: rational.h:59
avassert.h
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:180
av_cold
#define av_cold
Definition: attributes.h:90
av_dict_get
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:60
AVCodecContext::extradata_size
int extradata_size
Definition: avcodec.h:528
LibkvazaarContext::encoder
kvz_encoder * encoder
Definition: libkvazaar.c:47
AVDictionaryEntry::key
char * key
Definition: dict.h:90
AVCodecContext::ticks_per_frame
int ticks_per_frame
For some codecs, the time base is closer to the field rate than the frame rate.
Definition: avcodec.h:557
AV_CODEC_CAP_OTHER_THREADS
#define AV_CODEC_CAP_OTHER_THREADS
Codec supports multithreading through a method other than slice- or frame-level multithreading.
Definition: codec.h:121
init
int(* init)(AVBSFContext *ctx)
Definition: dts2pts_bsf.c:365
LibkvazaarContext::api
const kvz_api * api
Definition: libkvazaar.c:46
av_assert0
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
pix_fmts
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:296
ctx
AVFormatContext * ctx
Definition: movenc.c:48
AV_PIX_FMT_YUV420P
@ AV_PIX_FMT_YUV420P
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:66
CODEC_LONG_NAME
#define CODEC_LONG_NAME(str)
Definition: codec_internal.h:272
LIBAVUTIL_VERSION_INT
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
NULL
#define NULL
Definition: coverity.c:32
AVCodecContext::bit_rate
int64_t bit_rate
the average bitrate
Definition: avcodec.h:476
av_default_item_name
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:237
AV_PICTURE_TYPE_I
@ AV_PICTURE_TYPE_I
Intra.
Definition: avutil.h:274
error.h
AVCodecContext::time_base
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented.
Definition: avcodec.h:548
AV_CODEC_CAP_DR1
#define AV_CODEC_CAP_DR1
Codec uses get_buffer() or get_encode_buffer() for allocating buffers and supports custom allocators.
Definition: codec.h:52
codec_internal.h
AVPacket::dts
int64_t dts
Decompression timestamp in AVStream->time_base units; the time at which the packet is decompressed.
Definition: packet.h:373
AVERROR_EXTERNAL
#define AVERROR_EXTERNAL
Generic error in an external library.
Definition: error.h:59
attributes.h
AVPacket::flags
int flags
A combination of AV_PKT_FLAG values.
Definition: packet.h:380
av_dict_free
void av_dict_free(AVDictionary **pm)
Free all the memory allocated for an AVDictionary struct and all keys and values.
Definition: dict.c:223
LibkvazaarContext::config
kvz_config * config
Definition: libkvazaar.c:48
VE
#define VE
Definition: libkvazaar.c:302
log.h
AVPacket::pts
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: packet.h:367
AVCodecContext::extradata
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:527
internal.h
AV_CODEC_ID_HEVC
@ AV_CODEC_ID_HEVC
Definition: codec_id.h:226
libkvazaar_close
static av_cold int libkvazaar_close(AVCodecContext *avctx)
Definition: libkvazaar.c:151
av_mallocz
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:254
AVCodec::name
const char * name
Name of the codec implementation.
Definition: codec.h:191
AVCodecContext::height
int height
Definition: avcodec.h:598
AVCodecContext::pix_fmt
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
Definition: avcodec.h:635
avcodec.h
AVClass::class_name
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:71
frame
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
Definition: filter_design.txt:264
dict.h
AV_INPUT_BUFFER_PADDING_SIZE
#define AV_INPUT_BUFFER_PADDING_SIZE
Definition: defs.h:40
AVCodecContext
main external API structure.
Definition: avcodec.h:426
av_image_copy
void av_image_copy(uint8_t *dst_data[4], int dst_linesizes[4], const uint8_t *src_data[4], const int src_linesizes[4], enum AVPixelFormat pix_fmt, int width, int height)
Copy image in src_data to dst_data.
Definition: imgutils.c:422
AV_PICTURE_TYPE_B
@ AV_PICTURE_TYPE_B
Bi-dir predicted.
Definition: avutil.h:276
ff_get_encode_buffer
int ff_get_encode_buffer(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int flags)
Get a buffer for a packet.
Definition: encode.c:79
AVRational::den
int den
Denominator.
Definition: rational.h:60
AV_PIX_FMT_NONE
@ AV_PIX_FMT_NONE
Definition: pixfmt.h:65
av_dict_parse_string
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:200
AV_CODEC_CAP_DELAY
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: codec.h:76
AV_PICTURE_TYPE_P
@ AV_PICTURE_TYPE_P
Predicted.
Definition: avutil.h:275
AVMEDIA_TYPE_VIDEO
@ AVMEDIA_TYPE_VIDEO
Definition: avutil.h:201
mem.h
packet_internal.h
FF_CODEC_CAP_AUTO_THREADS
#define FF_CODEC_CAP_AUTO_THREADS
Codec handles avctx->thread_count == 0 (auto) internally.
Definition: codec_internal.h:73
AVDictionaryEntry
Definition: dict.h:89
AVPacket
This structure stores compressed data.
Definition: packet.h:351
AVCodecContext::priv_data
void * priv_data
Definition: avcodec.h:453
AVCodecContext::width
int width
picture width / height.
Definition: avcodec.h:598
imgutils.h
AVERROR_BUG
#define AVERROR_BUG
Internal bug, also see AVERROR_BUG2.
Definition: error.h:52
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
AVERROR_INVALIDDATA
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:61
AVDictionaryEntry::value
char * value
Definition: dict.h:91
AV_OPT_TYPE_STRING
@ AV_OPT_TYPE_STRING
Definition: opt.h:229
FF_QP2LAMBDA
#define FF_QP2LAMBDA
factor to convert from H.263 QP to lambda
Definition: avutil.h:227
AVCodecContext::sample_aspect_ratio
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:795
libkvazaar_encode
static int libkvazaar_encode(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: libkvazaar.c:163
av_get_pix_fmt_name
const char * av_get_pix_fmt_name(enum AVPixelFormat pix_fmt)
Return the short name for a pixel format, NULL in case pix_fmt is unknown.
Definition: pixdesc.c:2808