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 "internal.h"
40 #include "packet_internal.h"
41 
42 typedef struct LibkvazaarContext {
43  const AVClass *class;
44 
45  const kvz_api *api;
46  kvz_encoder *encoder;
47  kvz_config *config;
48 
49  char *kvz_params;
51 
53 {
54  LibkvazaarContext *const ctx = avctx->priv_data;
55  const kvz_api *const api = ctx->api = kvz_api_get(8);
56  kvz_config *cfg = NULL;
57  kvz_encoder *enc = NULL;
58 
59  /* Kvazaar requires width and height to be multiples of eight. */
60  if (avctx->width % 8 || avctx->height % 8) {
61  av_log(avctx, AV_LOG_ERROR,
62  "Video dimensions are not a multiple of 8 (%dx%d).\n",
63  avctx->width, avctx->height);
64  return AVERROR(ENOSYS);
65  }
66 
67  ctx->config = cfg = api->config_alloc();
68  if (!cfg) {
69  av_log(avctx, AV_LOG_ERROR,
70  "Could not allocate kvazaar config structure.\n");
71  return AVERROR(ENOMEM);
72  }
73 
74  if (!api->config_init(cfg)) {
75  av_log(avctx, AV_LOG_ERROR,
76  "Could not initialize kvazaar config structure.\n");
77  return AVERROR_BUG;
78  }
79 
80  cfg->width = avctx->width;
81  cfg->height = avctx->height;
82 
83  if (avctx->framerate.num > 0 && avctx->framerate.den > 0) {
84  cfg->framerate_num = avctx->framerate.num;
85  cfg->framerate_denom = avctx->framerate.den;
86  } else {
87  if (avctx->ticks_per_frame > INT_MAX / avctx->time_base.num) {
88  av_log(avctx, AV_LOG_ERROR,
89  "Could not set framerate for kvazaar: integer overflow\n");
90  return AVERROR(EINVAL);
91  }
92  cfg->framerate_num = avctx->time_base.den;
93  cfg->framerate_denom = avctx->time_base.num * avctx->ticks_per_frame;
94  }
95  cfg->target_bitrate = avctx->bit_rate;
96  cfg->vui.sar_width = avctx->sample_aspect_ratio.num;
97  cfg->vui.sar_height = avctx->sample_aspect_ratio.den;
98  if (avctx->bit_rate) {
99  cfg->rc_algorithm = KVZ_LAMBDA;
100  }
101 
102  if (ctx->kvz_params) {
103  AVDictionary *dict = NULL;
104  if (!av_dict_parse_string(&dict, ctx->kvz_params, "=", ",", 0)) {
105  AVDictionaryEntry *entry = NULL;
106  while ((entry = av_dict_get(dict, "", entry, AV_DICT_IGNORE_SUFFIX))) {
107  if (!api->config_parse(cfg, entry->key, entry->value)) {
108  av_log(avctx, AV_LOG_WARNING, "Invalid option: %s=%s.\n",
109  entry->key, entry->value);
110  }
111  }
112  }
113  av_dict_free(&dict);
114  }
115 
116  ctx->encoder = enc = api->encoder_open(cfg);
117  if (!enc) {
118  av_log(avctx, AV_LOG_ERROR, "Could not open kvazaar encoder.\n");
119  return AVERROR_BUG;
120  }
121 
122  if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
123  kvz_data_chunk *data_out = NULL;
124  kvz_data_chunk *chunk = NULL;
125  uint32_t len_out;
126  uint8_t *p;
127 
128  if (!api->encoder_headers(enc, &data_out, &len_out))
129  return AVERROR(ENOMEM);
130 
131  avctx->extradata = p = av_mallocz(len_out + AV_INPUT_BUFFER_PADDING_SIZE);
132  if (!p) {
133  ctx->api->chunk_free(data_out);
134  return AVERROR(ENOMEM);
135  }
136 
137  avctx->extradata_size = len_out;
138 
139  for (chunk = data_out; chunk != NULL; chunk = chunk->next) {
140  memcpy(p, chunk->data, chunk->len);
141  p += chunk->len;
142  }
143 
144  ctx->api->chunk_free(data_out);
145  }
146 
147  return 0;
148 }
149 
151 {
152  LibkvazaarContext *ctx = avctx->priv_data;
153 
154  if (ctx->api) {
155  ctx->api->encoder_close(ctx->encoder);
156  ctx->api->config_destroy(ctx->config);
157  }
158 
159  if (avctx->extradata)
160  av_freep(&avctx->extradata);
161 
162  return 0;
163 }
164 
166  AVPacket *avpkt,
167  const AVFrame *frame,
168  int *got_packet_ptr)
169 {
170  LibkvazaarContext *ctx = avctx->priv_data;
171  kvz_picture *input_pic = NULL;
172  kvz_picture *recon_pic = NULL;
173  kvz_frame_info frame_info;
174  kvz_data_chunk *data_out = NULL;
175  uint32_t len_out = 0;
176  int retval = 0;
177  int pict_type;
178 
179  *got_packet_ptr = 0;
180 
181  if (frame) {
182  if (frame->width != ctx->config->width ||
183  frame->height != ctx->config->height) {
184  av_log(avctx, AV_LOG_ERROR,
185  "Changing video dimensions during encoding is not supported. "
186  "(changed from %dx%d to %dx%d)\n",
187  ctx->config->width, ctx->config->height,
188  frame->width, frame->height);
189  retval = AVERROR_INVALIDDATA;
190  goto done;
191  }
192 
193  if (frame->format != avctx->pix_fmt) {
194  av_log(avctx, AV_LOG_ERROR,
195  "Changing pixel format during encoding is not supported. "
196  "(changed from %s to %s)\n",
198  av_get_pix_fmt_name(frame->format));
199  retval = AVERROR_INVALIDDATA;
200  goto done;
201  }
202 
203  // Allocate input picture for kvazaar.
204  input_pic = ctx->api->picture_alloc(frame->width, frame->height);
205  if (!input_pic) {
206  av_log(avctx, AV_LOG_ERROR, "Failed to allocate picture.\n");
207  retval = AVERROR(ENOMEM);
208  goto done;
209  }
210 
211  // Copy pixels from frame to input_pic.
212  {
213  int dst_linesizes[4] = {
214  frame->width,
215  frame->width / 2,
216  frame->width / 2,
217  0
218  };
219  av_image_copy(input_pic->data, dst_linesizes,
220  (const uint8_t **)frame->data, frame->linesize,
221  frame->format, frame->width, frame->height);
222  }
223 
224  input_pic->pts = frame->pts;
225  }
226 
227  retval = ctx->api->encoder_encode(ctx->encoder,
228  input_pic,
229  &data_out, &len_out,
230  &recon_pic, NULL,
231  &frame_info);
232  if (!retval) {
233  av_log(avctx, AV_LOG_ERROR, "Failed to encode frame.\n");
234  retval = AVERROR_INVALIDDATA;
235  goto done;
236  } else
237  retval = 0; /* kvazaar returns 1 on success */
238 
239  if (data_out) {
240  kvz_data_chunk *chunk = NULL;
241  uint64_t written = 0;
242 
243  retval = ff_alloc_packet2(avctx, avpkt, len_out, len_out);
244  if (retval < 0) {
245  av_log(avctx, AV_LOG_ERROR, "Failed to allocate output packet.\n");
246  goto done;
247  }
248 
249  for (chunk = data_out; chunk != NULL; chunk = chunk->next) {
250  av_assert0(written + chunk->len <= len_out);
251  memcpy(avpkt->data + written, chunk->data, chunk->len);
252  written += chunk->len;
253  }
254 
255  avpkt->pts = recon_pic->pts;
256  avpkt->dts = recon_pic->dts;
257  avpkt->flags = 0;
258  // IRAP VCL NAL unit types span the range
259  // [BLA_W_LP (16), RSV_IRAP_VCL23 (23)].
260  if (frame_info.nal_unit_type >= KVZ_NAL_BLA_W_LP &&
261  frame_info.nal_unit_type <= KVZ_NAL_RSV_IRAP_VCL23) {
262  avpkt->flags |= AV_PKT_FLAG_KEY;
263  }
264 
265  switch (frame_info.slice_type) {
266  case KVZ_SLICE_I:
267  pict_type = AV_PICTURE_TYPE_I;
268  break;
269  case KVZ_SLICE_P:
270  pict_type = AV_PICTURE_TYPE_P;
271  break;
272  case KVZ_SLICE_B:
273  pict_type = AV_PICTURE_TYPE_B;
274  break;
275  default:
276  av_log(avctx, AV_LOG_ERROR, "Unknown picture type encountered.\n");
277  return AVERROR_EXTERNAL;
278  }
279 #if FF_API_CODED_FRAME
281  avctx->coded_frame->pict_type = pict_type;
283 #endif
284 
285  ff_side_data_set_encoder_stats(avpkt, frame_info.qp * FF_QP2LAMBDA, NULL, 0, pict_type);
286 
287 #if FF_API_CODED_FRAME
289  avctx->coded_frame->quality = frame_info.qp * FF_QP2LAMBDA;
291 #endif
292 
293  *got_packet_ptr = 1;
294  }
295 
296 done:
297  ctx->api->picture_free(input_pic);
298  ctx->api->picture_free(recon_pic);
299  ctx->api->chunk_free(data_out);
300  return retval;
301 }
302 
303 static const enum AVPixelFormat pix_fmts[] = {
306 };
307 
308 #define OFFSET(x) offsetof(LibkvazaarContext, x)
309 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
310 static const AVOption options[] = {
311  { "kvazaar-params", "Set kvazaar parameters as a comma-separated list of key=value pairs.",
312  OFFSET(kvz_params), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
313  { NULL },
314 };
315 
316 static const AVClass class = {
317  .class_name = "libkvazaar",
318  .item_name = av_default_item_name,
319  .option = options,
321 };
322 
323 static const AVCodecDefault defaults[] = {
324  { "b", "0" },
325  { NULL },
326 };
327 
329  .name = "libkvazaar",
330  .long_name = NULL_IF_CONFIG_SMALL("libkvazaar H.265 / HEVC"),
331  .type = AVMEDIA_TYPE_VIDEO,
332  .id = AV_CODEC_ID_HEVC,
334  .pix_fmts = pix_fmts,
335 
336  .priv_class = &class,
337  .priv_data_size = sizeof(LibkvazaarContext),
338  .defaults = defaults,
339 
341  .encode2 = libkvazaar_encode,
342  .close = libkvazaar_close,
343 
345 
346  .wrapper_name = "libkvazaar",
347 };
#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:49
#define NULL
Definition: coverity.c:32
AVRational framerate
Definition: avcodec.h:2062
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
This structure describes decoded (raw) audio or video data.
Definition: frame.h:314
AVOption.
Definition: opt.h:248
int ff_side_data_set_encoder_stats(AVPacket *pkt, int quality, int64_t *error, int error_count, int pict_type)
Definition: avpacket.c:796
misc image utilities
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:200
int64_t bit_rate
the average bitrate
Definition: avcodec.h:581
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
Memory handling functions.
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:31
int num
Numerator.
Definition: rational.h:59
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:235
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:910
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
Definition: avcodec.h:741
static av_cold int libkvazaar_close(AVCodecContext *avctx)
Definition: libkvazaar.c:150
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:237
AVCodec ff_libkvazaar_encoder
Definition: libkvazaar.c:328
#define AV_CODEC_CAP_AUTO_THREADS
Codec supports avctx->thread_count == 0 (auto).
Definition: codec.h:118
#define VE
Definition: libkvazaar.c:309
AVCodec.
Definition: codec.h:190
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:654
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
#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:75
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
Public dictionary API.
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:33
#define OFFSET(x)
Definition: libkvazaar.c:308
#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:41
uint8_t
#define av_cold
Definition: attributes.h:88
AVOptions.
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:407
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:632
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:40
uint8_t * data
Definition: packet.h:363
kvz_encoder * encoder
Definition: libkvazaar.c:46
kvz_config * config
Definition: libkvazaar.c:47
#define av_log(a,...)
const kvz_api * api
Definition: libkvazaar.c:45
#define AV_PKT_FLAG_KEY
The packet contains a keyframe.
Definition: packet.h:401
int width
Definition: frame.h:372
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:194
error code definitions
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:115
void av_dict_free(AVDictionary **pm)
Free all the memory allocated for an AVDictionary struct and all keys and values. ...
Definition: dict.c:203
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:611
simple assert() macros that are a bit more flexible than ISO C assert().
const char * name
Name of the codec implementation.
Definition: codec.h:197
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
int flags
A combination of AV_PKT_FLAG values.
Definition: packet.h:369
static const AVOption options[]
Definition: libkvazaar.c:310
common internal API header
enum AVPictureType pict_type
Picture type of the frame.
Definition: frame.h:397
int width
picture width / height.
Definition: avcodec.h:704
AVFormatContext * ctx
Definition: movenc.c:48
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
int quality
quality (between 1 (good) and FF_LAMBDA_MAX (bad))
Definition: frame.h:437
int ticks_per_frame
For some codecs, the time base is closer to the field rate than the frame rate.
Definition: avcodec.h:663
static const AVCodecDefault defaults[]
Definition: libkvazaar.c:323
int format
format of the frame, -1 if unknown or unset Values correspond to enum AVPixelFormat for video frames...
Definition: frame.h:387
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:180
Libavcodec external API header.
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:345
main external API structure.
Definition: avcodec.h:531
int extradata_size
Definition: avcodec.h:633
#define AVERROR_BUG
Internal bug, also see AVERROR_BUG2.
Definition: error.h:50
Describe the class of an AVClass context structure.
Definition: log.h:67
static int libkvazaar_encode(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: libkvazaar.c:165
static av_cold int libkvazaar_init(AVCodecContext *avctx)
Definition: libkvazaar.c:52
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:303
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:328
#define AV_CODEC_FLAG_GLOBAL_HEADER
Place global headers in extradata instead of every keyframe.
Definition: avcodec.h:329
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:66
#define FF_DISABLE_DEPRECATION_WARNINGS
Definition: internal.h:83
common internal api header.
Bi-dir predicted.
Definition: avutil.h:276
attribute_deprecated AVFrame * coded_frame
the picture in the bitstream
Definition: avcodec.h:1759
char * key
Definition: dict.h:86
int den
Denominator.
Definition: rational.h:60
#define AV_INPUT_BUFFER_PADDING_SIZE
Required number of additionally allocated bytes at the end of the input bitstream for decoding...
Definition: avcodec.h:215
void * priv_data
Definition: avcodec.h:558
char * value
Definition: dict.h:87
#define FF_ENABLE_DEPRECATION_WARNINGS
Definition: internal.h:84
#define FF_QP2LAMBDA
factor to convert from H.263 QP to lambda
Definition: avutil.h:227
int64_t dts
Decompression timestamp in AVStream->time_base units; the time at which the packet is decompressed...
Definition: packet.h:362
int height
Definition: frame.h:372
#define av_freep(p)
#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:70
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:2489
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
#define AVERROR_EXTERNAL
Generic error in an external library.
Definition: error.h:57
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
This structure stores compressed data.
Definition: packet.h:340
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: packet.h:356
Predicted.
Definition: avutil.h:275