FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
frame.h
Go to the documentation of this file.
1 /*
2  *
3  * This file is part of FFmpeg.
4  *
5  * FFmpeg is free software; you can redistribute it and/or
6  * modify it under the terms of the GNU Lesser General Public
7  * License as published by the Free Software Foundation; either
8  * version 2.1 of the License, or (at your option) any later version.
9  *
10  * FFmpeg is distributed in the hope that it will be useful,
11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13  * Lesser General Public License for more details.
14  *
15  * You should have received a copy of the GNU Lesser General Public
16  * License along with FFmpeg; if not, write to the Free Software
17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18  */
19 
20 /**
21  * @file
22  * @ingroup lavu_frame
23  * reference-counted frame API
24  */
25 
26 #ifndef AVUTIL_FRAME_H
27 #define AVUTIL_FRAME_H
28 
29 #include <stdint.h>
30 
31 #include "avutil.h"
32 #include "buffer.h"
33 #include "dict.h"
34 #include "rational.h"
35 #include "samplefmt.h"
36 #include "pixfmt.h"
37 #include "version.h"
38 
39 
40 /**
41  * @defgroup lavu_frame AVFrame
42  * @ingroup lavu_data
43  *
44  * @{
45  * AVFrame is an abstraction for reference-counted raw multimedia data.
46  */
47 
49  /**
50  * The data is the AVPanScan struct defined in libavcodec.
51  */
53  /**
54  * ATSC A53 Part 4 Closed Captions.
55  * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data.
56  * The number of bytes of CC data is AVFrameSideData.size.
57  */
59  /**
60  * Stereoscopic 3d metadata.
61  * The data is the AVStereo3D struct defined in libavutil/stereo3d.h.
62  */
64  /**
65  * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h.
66  */
68  /**
69  * Metadata relevant to a downmix procedure.
70  * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h.
71  */
73  /**
74  * ReplayGain information in the form of the AVReplayGain struct.
75  */
77  /**
78  * This side data contains a 3x3 transformation matrix describing an affine
79  * transformation that needs to be applied to the frame for correct
80  * presentation.
81  *
82  * See libavutil/display.h for a detailed description of the data.
83  */
85  /**
86  * Active Format Description data consisting of a single byte as specified
87  * in ETSI TS 101 154 using AVActiveFormatDescription enum.
88  */
90  /**
91  * Motion vectors exported by some codecs (on demand through the export_mvs
92  * flag set in the libavcodec AVCodecContext flags2 option).
93  * The data is the AVMotionVector struct defined in
94  * libavutil/motion_vector.h.
95  */
97  /**
98  * Recommmends skipping the specified number of samples. This is exported
99  * only if the "skip_manual" AVOption is set in libavcodec.
100  * This has the same format as AV_PKT_DATA_SKIP_SAMPLES.
101  * @code
102  * u32le number of samples to skip from start of this packet
103  * u32le number of samples to skip from end of this packet
104  * u8 reason for start skip
105  * u8 reason for end skip (0=padding silence, 1=convergence)
106  * @endcode
107  */
109 
110  /**
111  * This side data must be associated with an audio frame and corresponds to
112  * enum AVAudioServiceType defined in avcodec.h.
113  */
115 };
116 
125 };
126 
127 
128 /**
129  * Structure to hold side data for an AVFrame.
130  *
131  * sizeof(AVFrameSideData) is not a part of the public ABI, so new fields may be added
132  * to the end with a minor bump.
133  */
134 typedef struct AVFrameSideData {
137  int size;
141 
142 /**
143  * This structure describes decoded (raw) audio or video data.
144  *
145  * AVFrame must be allocated using av_frame_alloc(). Note that this only
146  * allocates the AVFrame itself, the buffers for the data must be managed
147  * through other means (see below).
148  * AVFrame must be freed with av_frame_free().
149  *
150  * AVFrame is typically allocated once and then reused multiple times to hold
151  * different data (e.g. a single AVFrame to hold frames received from a
152  * decoder). In such a case, av_frame_unref() will free any references held by
153  * the frame and reset it to its original clean state before it
154  * is reused again.
155  *
156  * The data described by an AVFrame is usually reference counted through the
157  * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
158  * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
159  * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
160  * every single data plane must be contained in one of the buffers in
161  * AVFrame.buf or AVFrame.extended_buf.
162  * There may be a single buffer for all the data, or one separate buffer for
163  * each plane, or anything in between.
164  *
165  * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
166  * to the end with a minor bump.
167  * Similarly fields that are marked as to be only accessed by
168  * av_opt_ptr() can be reordered. This allows 2 forks to add fields
169  * without breaking compatibility with each other.
170  */
171 typedef struct AVFrame {
172 #define AV_NUM_DATA_POINTERS 8
173  /**
174  * pointer to the picture/channel planes.
175  * This might be different from the first allocated byte
176  *
177  * Some decoders access areas outside 0,0 - width,height, please
178  * see avcodec_align_dimensions2(). Some filters and swscale can read
179  * up to 16 bytes beyond the planes, if these filters are to be used,
180  * then 16 extra bytes must be allocated.
181  */
183 
184  /**
185  * For video, size in bytes of each picture line.
186  * For audio, size in bytes of each plane.
187  *
188  * For audio, only linesize[0] may be set. For planar audio, each channel
189  * plane must be the same size.
190  *
191  * For video the linesizes should be multiples of the CPUs alignment
192  * preference, this is 16 or 32 for modern desktop CPUs.
193  * Some code requires such alignment other code can be slower without
194  * correct alignment, for yet other it makes no difference.
195  *
196  * @note The linesize may be larger than the size of usable data -- there
197  * may be extra padding present for performance reasons.
198  */
200 
201  /**
202  * pointers to the data planes/channels.
203  *
204  * For video, this should simply point to data[].
205  *
206  * For planar audio, each channel has a separate data pointer, and
207  * linesize[0] contains the size of each channel buffer.
208  * For packed audio, there is just one data pointer, and linesize[0]
209  * contains the total size of the buffer for all channels.
210  *
211  * Note: Both data and extended_data should always be set in a valid frame,
212  * but for planar audio with more channels that can fit in data,
213  * extended_data must be used in order to access all channels.
214  */
216 
217  /**
218  * width and height of the video frame
219  */
220  int width, height;
221 
222  /**
223  * number of audio samples (per channel) described by this frame
224  */
226 
227  /**
228  * format of the frame, -1 if unknown or unset
229  * Values correspond to enum AVPixelFormat for video frames,
230  * enum AVSampleFormat for audio)
231  */
232  int format;
233 
234  /**
235  * 1 -> keyframe, 0-> not
236  */
238 
239  /**
240  * Picture type of the frame.
241  */
243 
244 #if FF_API_AVFRAME_LAVC
247 #endif
248 
249  /**
250  * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
251  */
253 
254  /**
255  * Presentation timestamp in time_base units (time when frame should be shown to user).
256  */
257  int64_t pts;
258 
259  /**
260  * PTS copied from the AVPacket that was decoded to produce this frame.
261  */
262  int64_t pkt_pts;
263 
264  /**
265  * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used)
266  * This is also the Presentation time of this AVFrame calculated from
267  * only AVPacket.dts values without pts values.
268  */
269  int64_t pkt_dts;
270 
271  /**
272  * picture number in bitstream order
273  */
275  /**
276  * picture number in display order
277  */
279 
280  /**
281  * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
282  */
283  int quality;
284 
285 #if FF_API_AVFRAME_LAVC
288 
289  /**
290  * QP table
291  */
293  int8_t *qscale_table;
294  /**
295  * QP store stride
296  */
298  int qstride;
299 
302 
303  /**
304  * mbskip_table[mb]>=1 if MB didn't change
305  * stride= mb_width = (width+15)>>4
306  */
309 
310  /**
311  * motion vector table
312  * @code
313  * example:
314  * int mv_sample_log2= 4 - motion_subsample_log2;
315  * int mb_width= (width+15)>>4;
316  * int mv_stride= (mb_width << mv_sample_log2) + 1;
317  * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
318  * @endcode
319  */
320  int16_t (*motion_val[2])[2];
321 
322  /**
323  * macroblock type table
324  * mb_type_base + mb_width + 2
325  */
327  uint32_t *mb_type;
328 
329  /**
330  * DCT coefficients
331  */
333  short *dct_coeff;
334 
335  /**
336  * motion reference frame index
337  * the order in which these are stored can depend on the codec.
338  */
340  int8_t *ref_index[2];
341 #endif
342 
343  /**
344  * for some private data of the user
345  */
346  void *opaque;
347 
348  /**
349  * error
350  */
352 
353 #if FF_API_AVFRAME_LAVC
355  int type;
356 #endif
357 
358  /**
359  * When decoding, this signals how much the picture must be delayed.
360  * extra_delay = repeat_pict / (2*fps)
361  */
363 
364  /**
365  * The content of the picture is interlaced.
366  */
368 
369  /**
370  * If the content is interlaced, is top field displayed first.
371  */
373 
374  /**
375  * Tell user application that palette has changed from previous frame.
376  */
378 
379 #if FF_API_AVFRAME_LAVC
382 
383  /**
384  * Pan scan.
385  */
388 #endif
389 
390  /**
391  * reordered opaque 64bit (generally an integer or a double precision float
392  * PTS but can be anything).
393  * The user sets AVCodecContext.reordered_opaque to represent the input at
394  * that time,
395  * the decoder reorders values as needed and sets AVFrame.reordered_opaque
396  * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
397  * @deprecated in favor of pkt_pts
398  */
400 
401 #if FF_API_AVFRAME_LAVC
402  /**
403  * @deprecated this field is unused
404  */
406 
411 
412  /**
413  * log2 of the size of the block which a single vector in motion_val represents:
414  * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
415  */
417 #endif
418 
419  /**
420  * Sample rate of the audio data.
421  */
423 
424  /**
425  * Channel layout of the audio data.
426  */
427  uint64_t channel_layout;
428 
429  /**
430  * AVBuffer references backing the data for this frame. If all elements of
431  * this array are NULL, then this frame is not reference counted. This array
432  * must be filled contiguously -- if buf[i] is non-NULL then buf[j] must
433  * also be non-NULL for all j < i.
434  *
435  * There may be at most one AVBuffer per data plane, so for video this array
436  * always contains all the references. For planar audio with more than
437  * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
438  * this array. Then the extra AVBufferRef pointers are stored in the
439  * extended_buf array.
440  */
442 
443  /**
444  * For planar audio which requires more than AV_NUM_DATA_POINTERS
445  * AVBufferRef pointers, this array will hold all the references which
446  * cannot fit into AVFrame.buf.
447  *
448  * Note that this is different from AVFrame.extended_data, which always
449  * contains all the pointers. This array only contains the extra pointers,
450  * which cannot fit into AVFrame.buf.
451  *
452  * This array is always allocated using av_malloc() by whoever constructs
453  * the frame. It is freed in av_frame_unref().
454  */
456  /**
457  * Number of elements in extended_buf.
458  */
460 
463 
464 /**
465  * @defgroup lavu_frame_flags AV_FRAME_FLAGS
466  * Flags describing additional frame properties.
467  *
468  * @{
469  */
470 
471 /**
472  * The frame data may be corrupted, e.g. due to decoding errors.
473  */
474 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
475 /**
476  * @}
477  */
478 
479  /**
480  * Frame flags, a combination of @ref lavu_frame_flags
481  */
482  int flags;
483 
484  /**
485  * MPEG vs JPEG YUV range.
486  * It must be accessed using av_frame_get_color_range() and
487  * av_frame_set_color_range().
488  * - encoding: Set by user
489  * - decoding: Set by libavcodec
490  */
492 
494 
496 
497  /**
498  * YUV colorspace type.
499  * It must be accessed using av_frame_get_colorspace() and
500  * av_frame_set_colorspace().
501  * - encoding: Set by user
502  * - decoding: Set by libavcodec
503  */
505 
507 
508  /**
509  * frame timestamp estimated using various heuristics, in stream time base
510  * Code outside libavcodec should access this field using:
511  * av_frame_get_best_effort_timestamp(frame)
512  * - encoding: unused
513  * - decoding: set by libavcodec, read by user.
514  */
516 
517  /**
518  * reordered pos from the last AVPacket that has been input into the decoder
519  * Code outside libavcodec should access this field using:
520  * av_frame_get_pkt_pos(frame)
521  * - encoding: unused
522  * - decoding: Read by user.
523  */
524  int64_t pkt_pos;
525 
526  /**
527  * duration of the corresponding packet, expressed in
528  * AVStream->time_base units, 0 if unknown.
529  * Code outside libavcodec should access this field using:
530  * av_frame_get_pkt_duration(frame)
531  * - encoding: unused
532  * - decoding: Read by user.
533  */
534  int64_t pkt_duration;
535 
536  /**
537  * metadata.
538  * Code outside libavcodec should access this field using:
539  * av_frame_get_metadata(frame)
540  * - encoding: Set by user.
541  * - decoding: Set by libavcodec.
542  */
544 
545  /**
546  * decode error flags of the frame, set to a combination of
547  * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
548  * were errors during the decoding.
549  * Code outside libavcodec should access this field using:
550  * av_frame_get_decode_error_flags(frame)
551  * - encoding: unused
552  * - decoding: set by libavcodec, read by user.
553  */
555 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1
556 #define FF_DECODE_ERROR_MISSING_REFERENCE 2
557 
558  /**
559  * number of audio channels, only used for audio.
560  * Code outside libavcodec should access this field using:
561  * av_frame_get_channels(frame)
562  * - encoding: unused
563  * - decoding: Read by user.
564  */
565  int channels;
566 
567  /**
568  * size of the corresponding packet containing the compressed
569  * frame. It must be accessed using av_frame_get_pkt_size() and
570  * av_frame_set_pkt_size().
571  * It is set to a negative value if unknown.
572  * - encoding: unused
573  * - decoding: set by libavcodec, read by user.
574  */
575  int pkt_size;
576 
577  /**
578  * Not to be accessed directly from outside libavutil
579  */
581 } AVFrame;
582 
583 /**
584  * Accessors for some AVFrame fields.
585  * The position of these field in the structure is not part of the ABI,
586  * they should not be accessed directly outside libavcodec.
587  */
590 int64_t av_frame_get_pkt_duration (const AVFrame *frame);
591 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val);
592 int64_t av_frame_get_pkt_pos (const AVFrame *frame);
593 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val);
594 int64_t av_frame_get_channel_layout (const AVFrame *frame);
596 int av_frame_get_channels (const AVFrame *frame);
607 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
613 
614 /**
615  * Get the name of a colorspace.
616  * @return a static string identifying the colorspace; can be NULL.
617  */
618 const char *av_get_colorspace_name(enum AVColorSpace val);
619 
620 /**
621  * Allocate an AVFrame and set its fields to default values. The resulting
622  * struct must be freed using av_frame_free().
623  *
624  * @return An AVFrame filled with default values or NULL on failure.
625  *
626  * @note this only allocates the AVFrame itself, not the data buffers. Those
627  * must be allocated through other means, e.g. with av_frame_get_buffer() or
628  * manually.
629  */
630 AVFrame *av_frame_alloc(void);
631 
632 /**
633  * Free the frame and any dynamically allocated objects in it,
634  * e.g. extended_data. If the frame is reference counted, it will be
635  * unreferenced first.
636  *
637  * @param frame frame to be freed. The pointer will be set to NULL.
638  */
639 void av_frame_free(AVFrame **frame);
640 
641 /**
642  * Set up a new reference to the data described by the source frame.
643  *
644  * Copy frame properties from src to dst and create a new reference for each
645  * AVBufferRef from src.
646  *
647  * If src is not reference counted, new buffers are allocated and the data is
648  * copied.
649  *
650  * @return 0 on success, a negative AVERROR on error
651  */
652 int av_frame_ref(AVFrame *dst, const AVFrame *src);
653 
654 /**
655  * Create a new frame that references the same data as src.
656  *
657  * This is a shortcut for av_frame_alloc()+av_frame_ref().
658  *
659  * @return newly created AVFrame on success, NULL on error.
660  */
662 
663 /**
664  * Unreference all the buffers referenced by frame and reset the frame fields.
665  */
667 
668 /**
669  * Move everything contained in src to dst and reset src.
670  */
671 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
672 
673 /**
674  * Allocate new buffer(s) for audio or video data.
675  *
676  * The following fields must be set on frame before calling this function:
677  * - format (pixel format for video, sample format for audio)
678  * - width and height for video
679  * - nb_samples and channel_layout for audio
680  *
681  * This function will fill AVFrame.data and AVFrame.buf arrays and, if
682  * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
683  * For planar formats, one buffer will be allocated for each plane.
684  *
685  * @param frame frame in which to store the new buffers.
686  * @param align required buffer size alignment
687  *
688  * @return 0 on success, a negative AVERROR on error.
689  */
691 
692 /**
693  * Check if the frame data is writable.
694  *
695  * @return A positive value if the frame data is writable (which is true if and
696  * only if each of the underlying buffers has only one reference, namely the one
697  * stored in this frame). Return 0 otherwise.
698  *
699  * If 1 is returned the answer is valid until av_buffer_ref() is called on any
700  * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
701  *
702  * @see av_frame_make_writable(), av_buffer_is_writable()
703  */
705 
706 /**
707  * Ensure that the frame data is writable, avoiding data copy if possible.
708  *
709  * Do nothing if the frame is writable, allocate new buffers and copy the data
710  * if it is not.
711  *
712  * @return 0 on success, a negative AVERROR on error.
713  *
714  * @see av_frame_is_writable(), av_buffer_is_writable(),
715  * av_buffer_make_writable()
716  */
718 
719 /**
720  * Copy the frame data from src to dst.
721  *
722  * This function does not allocate anything, dst must be already initialized and
723  * allocated with the same parameters as src.
724  *
725  * This function only copies the frame data (i.e. the contents of the data /
726  * extended data arrays), not any other properties.
727  *
728  * @return >= 0 on success, a negative AVERROR on error.
729  */
730 int av_frame_copy(AVFrame *dst, const AVFrame *src);
731 
732 /**
733  * Copy only "metadata" fields from src to dst.
734  *
735  * Metadata for the purpose of this function are those fields that do not affect
736  * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
737  * aspect ratio (for video), but not width/height or channel layout.
738  * Side data is also copied.
739  */
740 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
741 
742 /**
743  * Get the buffer reference a given data plane is stored in.
744  *
745  * @param plane index of the data plane of interest in frame->extended_data.
746  *
747  * @return the buffer reference that contains the plane or NULL if the input
748  * frame is not valid.
749  */
751 
752 /**
753  * Add a new side data to a frame.
754  *
755  * @param frame a frame to which the side data should be added
756  * @param type type of the added side data
757  * @param size size of the side data
758  *
759  * @return newly added side data on success, NULL on error
760  */
763  int size);
764 
765 /**
766  * @return a pointer to the side data of a given type on success, NULL if there
767  * is no side data with such type in this frame.
768  */
771 
772 /**
773  * If side data of the supplied type exists in the frame, free it and remove it
774  * from the frame.
775  */
777 
778 /**
779  * @return a string identifying the side data type
780  */
782 
783 /**
784  * @}
785  */
786 
787 #endif /* AVUTIL_FRAME_H */