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 
107 };
108 
109 typedef struct AVFrameSideData {
112  int size;
115 
116 /**
117  * This structure describes decoded (raw) audio or video data.
118  *
119  * AVFrame must be allocated using av_frame_alloc(). Note that this only
120  * allocates the AVFrame itself, the buffers for the data must be managed
121  * through other means (see below).
122  * AVFrame must be freed with av_frame_free().
123  *
124  * AVFrame is typically allocated once and then reused multiple times to hold
125  * different data (e.g. a single AVFrame to hold frames received from a
126  * decoder). In such a case, av_frame_unref() will free any references held by
127  * the frame and reset it to its original clean state before it
128  * is reused again.
129  *
130  * The data described by an AVFrame is usually reference counted through the
131  * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
132  * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
133  * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
134  * every single data plane must be contained in one of the buffers in
135  * AVFrame.buf or AVFrame.extended_buf.
136  * There may be a single buffer for all the data, or one separate buffer for
137  * each plane, or anything in between.
138  *
139  * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
140  * to the end with a minor bump.
141  * Similarly fields that are marked as to be only accessed by
142  * av_opt_ptr() can be reordered. This allows 2 forks to add fields
143  * without breaking compatibility with each other.
144  */
145 typedef struct AVFrame {
146 #define AV_NUM_DATA_POINTERS 8
147  /**
148  * pointer to the picture/channel planes.
149  * This might be different from the first allocated byte
150  *
151  * Some decoders access areas outside 0,0 - width,height, please
152  * see avcodec_align_dimensions2(). Some filters and swscale can read
153  * up to 16 bytes beyond the planes, if these filters are to be used,
154  * then 16 extra bytes must be allocated.
155  */
157 
158  /**
159  * For video, size in bytes of each picture line.
160  * For audio, size in bytes of each plane.
161  *
162  * For audio, only linesize[0] may be set. For planar audio, each channel
163  * plane must be the same size.
164  *
165  * For video the linesizes should be multiplies of the CPUs alignment
166  * preference, this is 16 or 32 for modern desktop CPUs.
167  * Some code requires such alignment other code can be slower without
168  * correct alignment, for yet other it makes no difference.
169  *
170  * @note The linesize may be larger than the size of usable data -- there
171  * may be extra padding present for performance reasons.
172  */
174 
175  /**
176  * pointers to the data planes/channels.
177  *
178  * For video, this should simply point to data[].
179  *
180  * For planar audio, each channel has a separate data pointer, and
181  * linesize[0] contains the size of each channel buffer.
182  * For packed audio, there is just one data pointer, and linesize[0]
183  * contains the total size of the buffer for all channels.
184  *
185  * Note: Both data and extended_data should always be set in a valid frame,
186  * but for planar audio with more channels that can fit in data,
187  * extended_data must be used in order to access all channels.
188  */
190 
191  /**
192  * width and height of the video frame
193  */
194  int width, height;
195 
196  /**
197  * number of audio samples (per channel) described by this frame
198  */
200 
201  /**
202  * format of the frame, -1 if unknown or unset
203  * Values correspond to enum AVPixelFormat for video frames,
204  * enum AVSampleFormat for audio)
205  */
206  int format;
207 
208  /**
209  * 1 -> keyframe, 0-> not
210  */
212 
213  /**
214  * Picture type of the frame.
215  */
217 
218 #if FF_API_AVFRAME_LAVC
221 #endif
222 
223  /**
224  * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
225  */
227 
228  /**
229  * Presentation timestamp in time_base units (time when frame should be shown to user).
230  */
231  int64_t pts;
232 
233  /**
234  * PTS copied from the AVPacket that was decoded to produce this frame.
235  */
236  int64_t pkt_pts;
237 
238  /**
239  * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used)
240  * This is also the Presentation time of this AVFrame calculated from
241  * only AVPacket.dts values without pts values.
242  */
243  int64_t pkt_dts;
244 
245  /**
246  * picture number in bitstream order
247  */
249  /**
250  * picture number in display order
251  */
253 
254  /**
255  * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
256  */
257  int quality;
258 
259 #if FF_API_AVFRAME_LAVC
262 
263  /**
264  * QP table
265  */
267  int8_t *qscale_table;
268  /**
269  * QP store stride
270  */
272  int qstride;
273 
276 
277  /**
278  * mbskip_table[mb]>=1 if MB didn't change
279  * stride= mb_width = (width+15)>>4
280  */
283 
284  /**
285  * motion vector table
286  * @code
287  * example:
288  * int mv_sample_log2= 4 - motion_subsample_log2;
289  * int mb_width= (width+15)>>4;
290  * int mv_stride= (mb_width << mv_sample_log2) + 1;
291  * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
292  * @endcode
293  */
294  int16_t (*motion_val[2])[2];
295 
296  /**
297  * macroblock type table
298  * mb_type_base + mb_width + 2
299  */
301  uint32_t *mb_type;
302 
303  /**
304  * DCT coefficients
305  */
307  short *dct_coeff;
308 
309  /**
310  * motion reference frame index
311  * the order in which these are stored can depend on the codec.
312  */
314  int8_t *ref_index[2];
315 #endif
316 
317  /**
318  * for some private data of the user
319  */
320  void *opaque;
321 
322  /**
323  * error
324  */
326 
327 #if FF_API_AVFRAME_LAVC
329  int type;
330 #endif
331 
332  /**
333  * When decoding, this signals how much the picture must be delayed.
334  * extra_delay = repeat_pict / (2*fps)
335  */
337 
338  /**
339  * The content of the picture is interlaced.
340  */
342 
343  /**
344  * If the content is interlaced, is top field displayed first.
345  */
347 
348  /**
349  * Tell user application that palette has changed from previous frame.
350  */
352 
353 #if FF_API_AVFRAME_LAVC
356 
357  /**
358  * Pan scan.
359  */
362 #endif
363 
364  /**
365  * reordered opaque 64bit (generally an integer or a double precision float
366  * PTS but can be anything).
367  * The user sets AVCodecContext.reordered_opaque to represent the input at
368  * that time,
369  * the decoder reorders values as needed and sets AVFrame.reordered_opaque
370  * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
371  * @deprecated in favor of pkt_pts
372  */
374 
375 #if FF_API_AVFRAME_LAVC
376  /**
377  * @deprecated this field is unused
378  */
380 
385 
386  /**
387  * log2 of the size of the block which a single vector in motion_val represents:
388  * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
389  */
391 #endif
392 
393  /**
394  * Sample rate of the audio data.
395  */
397 
398  /**
399  * Channel layout of the audio data.
400  */
401  uint64_t channel_layout;
402 
403  /**
404  * AVBuffer references backing the data for this frame. If all elements of
405  * this array are NULL, then this frame is not reference counted.
406  *
407  * There may be at most one AVBuffer per data plane, so for video this array
408  * always contains all the references. For planar audio with more than
409  * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
410  * this array. Then the extra AVBufferRef pointers are stored in the
411  * extended_buf array.
412  */
414 
415  /**
416  * For planar audio which requires more than AV_NUM_DATA_POINTERS
417  * AVBufferRef pointers, this array will hold all the references which
418  * cannot fit into AVFrame.buf.
419  *
420  * Note that this is different from AVFrame.extended_data, which always
421  * contains all the pointers. This array only contains the extra pointers,
422  * which cannot fit into AVFrame.buf.
423  *
424  * This array is always allocated using av_malloc() by whoever constructs
425  * the frame. It is freed in av_frame_unref().
426  */
428  /**
429  * Number of elements in extended_buf.
430  */
432 
435 
436 /**
437  * @defgroup lavu_frame_flags AV_FRAME_FLAGS
438  * Flags describing additional frame properties.
439  *
440  * @{
441  */
442 
443 /**
444  * The frame data may be corrupted, e.g. due to decoding errors.
445  */
446 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
447 /**
448  * @}
449  */
450 
451  /**
452  * Frame flags, a combination of @ref lavu_frame_flags
453  */
454  int flags;
455 
456  /**
457  * MPEG vs JPEG YUV range.
458  * It must be accessed using av_frame_get_color_range() and
459  * av_frame_set_color_range().
460  * - encoding: Set by user
461  * - decoding: Set by libavcodec
462  */
464 
466 
468 
469  /**
470  * YUV colorspace type.
471  * It must be accessed using av_frame_get_colorspace() and
472  * av_frame_set_colorspace().
473  * - encoding: Set by user
474  * - decoding: Set by libavcodec
475  */
477 
479 
480  /**
481  * frame timestamp estimated using various heuristics, in stream time base
482  * Code outside libavcodec should access this field using:
483  * av_frame_get_best_effort_timestamp(frame)
484  * - encoding: unused
485  * - decoding: set by libavcodec, read by user.
486  */
488 
489  /**
490  * reordered pos from the last AVPacket that has been input into the decoder
491  * Code outside libavcodec should access this field using:
492  * av_frame_get_pkt_pos(frame)
493  * - encoding: unused
494  * - decoding: Read by user.
495  */
496  int64_t pkt_pos;
497 
498  /**
499  * duration of the corresponding packet, expressed in
500  * AVStream->time_base units, 0 if unknown.
501  * Code outside libavcodec should access this field using:
502  * av_frame_get_pkt_duration(frame)
503  * - encoding: unused
504  * - decoding: Read by user.
505  */
506  int64_t pkt_duration;
507 
508  /**
509  * metadata.
510  * Code outside libavcodec should access this field using:
511  * av_frame_get_metadata(frame)
512  * - encoding: Set by user.
513  * - decoding: Set by libavcodec.
514  */
516 
517  /**
518  * decode error flags of the frame, set to a combination of
519  * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
520  * were errors during the decoding.
521  * Code outside libavcodec should access this field using:
522  * av_frame_get_decode_error_flags(frame)
523  * - encoding: unused
524  * - decoding: set by libavcodec, read by user.
525  */
527 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1
528 #define FF_DECODE_ERROR_MISSING_REFERENCE 2
529 
530  /**
531  * number of audio channels, only used for audio.
532  * Code outside libavcodec should access this field using:
533  * av_frame_get_channels(frame)
534  * - encoding: unused
535  * - decoding: Read by user.
536  */
537  int channels;
538 
539  /**
540  * size of the corresponding packet containing the compressed
541  * frame. It must be accessed using av_frame_get_pkt_size() and
542  * av_frame_set_pkt_size().
543  * It is set to a negative value if unknown.
544  * - encoding: unused
545  * - decoding: set by libavcodec, read by user.
546  */
547  int pkt_size;
548 
549  /**
550  * Not to be accessed directly from outside libavutil
551  */
553 } AVFrame;
554 
555 /**
556  * Accessors for some AVFrame fields.
557  * The position of these field in the structure is not part of the ABI,
558  * they should not be accessed directly outside libavcodec.
559  */
562 int64_t av_frame_get_pkt_duration (const AVFrame *frame);
563 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val);
564 int64_t av_frame_get_pkt_pos (const AVFrame *frame);
565 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val);
566 int64_t av_frame_get_channel_layout (const AVFrame *frame);
568 int av_frame_get_channels (const AVFrame *frame);
579 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
585 
586 /**
587  * Get the name of a colorspace.
588  * @return a static string identifying the colorspace; can be NULL.
589  */
590 const char *av_get_colorspace_name(enum AVColorSpace val);
591 
592 /**
593  * Allocate an AVFrame and set its fields to default values. The resulting
594  * struct must be freed using av_frame_free().
595  *
596  * @return An AVFrame filled with default values or NULL on failure.
597  *
598  * @note this only allocates the AVFrame itself, not the data buffers. Those
599  * must be allocated through other means, e.g. with av_frame_get_buffer() or
600  * manually.
601  */
602 AVFrame *av_frame_alloc(void);
603 
604 /**
605  * Free the frame and any dynamically allocated objects in it,
606  * e.g. extended_data. If the frame is reference counted, it will be
607  * unreferenced first.
608  *
609  * @param frame frame to be freed. The pointer will be set to NULL.
610  */
611 void av_frame_free(AVFrame **frame);
612 
613 /**
614  * Set up a new reference to the data described by the source frame.
615  *
616  * Copy frame properties from src to dst and create a new reference for each
617  * AVBufferRef from src.
618  *
619  * If src is not reference counted, new buffers are allocated and the data is
620  * copied.
621  *
622  * @return 0 on success, a negative AVERROR on error
623  */
624 int av_frame_ref(AVFrame *dst, const AVFrame *src);
625 
626 /**
627  * Create a new frame that references the same data as src.
628  *
629  * This is a shortcut for av_frame_alloc()+av_frame_ref().
630  *
631  * @return newly created AVFrame on success, NULL on error.
632  */
634 
635 /**
636  * Unreference all the buffers referenced by frame and reset the frame fields.
637  */
639 
640 /**
641  * Move everythnig contained in src to dst and reset src.
642  */
643 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
644 
645 /**
646  * Allocate new buffer(s) for audio or video data.
647  *
648  * The following fields must be set on frame before calling this function:
649  * - format (pixel format for video, sample format for audio)
650  * - width and height for video
651  * - nb_samples and channel_layout for audio
652  *
653  * This function will fill AVFrame.data and AVFrame.buf arrays and, if
654  * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
655  * For planar formats, one buffer will be allocated for each plane.
656  *
657  * @param frame frame in which to store the new buffers.
658  * @param align required buffer size alignment
659  *
660  * @return 0 on success, a negative AVERROR on error.
661  */
663 
664 /**
665  * Check if the frame data is writable.
666  *
667  * @return A positive value if the frame data is writable (which is true if and
668  * only if each of the underlying buffers has only one reference, namely the one
669  * stored in this frame). Return 0 otherwise.
670  *
671  * If 1 is returned the answer is valid until av_buffer_ref() is called on any
672  * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
673  *
674  * @see av_frame_make_writable(), av_buffer_is_writable()
675  */
677 
678 /**
679  * Ensure that the frame data is writable, avoiding data copy if possible.
680  *
681  * Do nothing if the frame is writable, allocate new buffers and copy the data
682  * if it is not.
683  *
684  * @return 0 on success, a negative AVERROR on error.
685  *
686  * @see av_frame_is_writable(), av_buffer_is_writable(),
687  * av_buffer_make_writable()
688  */
690 
691 /**
692  * Copy the frame data from src to dst.
693  *
694  * This function does not allocate anything, dst must be already initialized and
695  * allocated with the same parameters as src.
696  *
697  * This function only copies the frame data (i.e. the contents of the data /
698  * extended data arrays), not any other properties.
699  *
700  * @return >= 0 on success, a negative AVERROR on error.
701  */
702 int av_frame_copy(AVFrame *dst, const AVFrame *src);
703 
704 /**
705  * Copy only "metadata" fields from src to dst.
706  *
707  * Metadata for the purpose of this function are those fields that do not affect
708  * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
709  * aspect ratio (for video), but not width/height or channel layout.
710  * Side data is also copied.
711  */
712 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
713 
714 /**
715  * Get the buffer reference a given data plane is stored in.
716  *
717  * @param plane index of the data plane of interest in frame->extended_data.
718  *
719  * @return the buffer reference that contains the plane or NULL if the input
720  * frame is not valid.
721  */
723 
724 /**
725  * Add a new side data to a frame.
726  *
727  * @param frame a frame to which the side data should be added
728  * @param type type of the added side data
729  * @param size size of the side data
730  *
731  * @return newly added side data on success, NULL on error
732  */
735  int size);
736 
737 /**
738  * @return a pointer to the side data of a given type on success, NULL if there
739  * is no side data with such type in this frame.
740  */
743 
744 /**
745  * If side data of the supplied type exists in the frame, free it and remove it
746  * from the frame.
747  */
749 
750 /**
751  * @return a string identifying the side data type
752  */
754 
755 /**
756  * @}
757  */
758 
759 #endif /* AVUTIL_FRAME_H */