4 This document describes the design for arbitrary per-buffer metadata.
6 Buffer metadata typically describes the lowlevel properties of the buffer
7 content. These properties are typically not negotiated with caps but they are
8 negotiated in the bufferpools.
10 Some examples of metadata:
14 - interlacing information
15 - video alignment, cropping, panning information
16 - extra container information such as granulepos, ...
17 - extra global buffer properties
24 * allocation, free, low fragmentation
25 * access to the metadata fields, preferably not much slower than directly
26 accessing a C structure field
27 - It must be extensible. Elements should be able to add new arbitrary metadata
28 without requiring much effort. Also new metadata fields should not break API
30 - It plays nice with subbuffers. When a subbuffer is created, the various
31 buffer metadata should be copied/updated correctly.
32 - We should be able to negotiate metadata between elements
39 Video data is sometimes allocated in non-contiguous planes for the Y and the UV
40 data. We need to be able to specify the data on a buffer using multiple
41 pointers in memory. We also need to be able to specify the stride for these
46 Some elements might need to store extra data for a buffer. This is typically
47 done when the resources are allocated from another subsystem such as OMX or
50 * Processing information
52 Pan and crop information can be added to the buffer data when the downstream
53 element can understand and use this metadata. An imagesink can, for example,
54 use the pan and cropping formation when it blits the image on the screen
61 A GstMeta is a structure as follows:
64 const GstMetaInfo *info; /* tag and info for the meta item */
67 The purpose of the this structure is to serve as a common header for all metadata
68 information that we can attach to a buffer. Specific metadata, such as timing metadata,
69 will have this structure as the first field. For example:
71 struct _GstMetaTiming {
72 GstMeta meta; /* common meta header */
74 GstClockTime dts; /* decoding timestamp */
75 GstClockTime pts; /* presentation timestamp */
76 GstClockTime duration; /* duration of the data */
77 GstClockTime clock_rate; /* clock rate for the above values */
80 Or another example for the video memory regions that consists of both fields and
84 #define GST_VIDEO_MAX_PLANES 4
92 GstVideoFormat format;
97 gsize offset[GST_VIDEO_MAX_PLANES]; /* offset in the buffer memory region of the
99 gint stride[GST_VIDEO_MAX_PLANES]; /* stride of the image lines. Can be negative when
100 * the image is upside-down */
102 gpointer (*map) (GstMetaVideo *meta, guint plane, gint *stride,
104 gboolean (*unmap) (GstMetaVideo *meta, guint plane, gpointer data);
107 gpointer gst_meta_video_map (GstMetaVideo *meta, guint plane, gint *stride,
109 gboolean gst_meta_video_unmap (GstMetaVideo *meta, guint plane, gpointer data);
111 GstMeta derived structures define the API of the metadata. The API can consist of
112 fields and/or methods. It is possible to have different implementations for the
113 same GstMeta structure.
115 The implementation of the GstMeta api would typically add more fields to the
116 public structure that allow it to implement the API.
118 GstMetaInfo will point to more information about the metadata and looks like this:
120 struct _GstMetaInfo {
121 GQuark api; /* api name */
122 GType type; /* implementation type */
123 gsize size; /* size of the structure */
125 GstMetaInitFunction init_func;
126 GstMetaFreeFunction free_func;
127 GstMetaCopyFunction copy_func;
128 GstMetaTransformFunction transform_func;
131 api will contain a GQuark of the metadata api. A repository of registered MetaInfo
132 will be maintained by the core. We will register some common metadata structures
133 in core and some media specific info for audio/video/text in -base. Plugins can
134 register additional custom metadata.
136 For each implementation of api, there will thus be a unique GstMetaInfo. In the
137 case of metadata with a well defined API, the implementation specific init
138 function will setup the methods in the metadata structure. A unique GType will
139 be made for each implementation and stored in the type field.
141 Along with the metadata description we will have functions to initialize/free (and/or refcount)
142 a specific GstMeta instance. We also have the possibility to add a custom
143 transform function that can be used to modify the metadata when a transformation
146 There are no explicit methods to serialize and deserialize the metadata. Since
147 each type has a GType, we can reuse the GValue transform functions for this.
149 The purpose of the separate MetaInfo is to not have to carry the free/init functions in
150 each buffer instance but to define them globally. We still want quick access to the info
151 so we need to make the buffer metadata point to the info.
153 Technically we could also specify the field and types in the MetaInfo and
154 provide a generic API to retrieve the metadata fields without the need for a
155 header file. We will not do this yet.
157 Allocation of the GstBuffer structure will result in the allocation of a memory region
158 of a customizable size (512 bytes). Only the first sizeof (GstBuffer) bytes of this
159 region will initially be used. The remaining bytes will be part of the free metadata
160 region of the buffer. Different implementations are possible and are invisible
163 The complete buffer with metadata could, for example, look as follows:
165 +-------------------------------------+
166 GstMiniObject | GType (GstBuffer) |
167 | refcount, flags, copy/disp/free |
168 +-------------------------------------+
169 GstBuffer | caps, parent, pool |
170 +.....................................+
172 +- | info ------> GstMetaInfo
173 GstMetaTiming | | | |
178 + . . . . . . . . . . . . . . . . . . + |
180 GstMetaVideo +- +- | info ------> GstMetaInfo
188 | | private fields | |
189 GstMetaVideoImpl | | ... | |
192 + . . . . . . . . . . . . . . . . . . + .
199 Buffers are created using the normal gst_buffer_new functions. The standard fields
200 are initialized as usual. A memory area that is bigger than the structure size
201 is allocated for the buffer metadata.
205 After creating a buffer, the application can set caps and add metadata
208 To add or retrieve metadata, a handle to a GstMetaInfo structure needs to be
209 obtained. This defines the implementation and API of the metadata. Usually, a
210 handle to this info structure can be obtained by calling a public _get_info()
211 method from a shared library (for shared metadata).
213 The following defines can usually be found in the shared .h file.
215 GstMetaInfo * gst_meta_timing_get_info();
216 #define GST_META_TIMING_INFO (gst_meta_timing_get_info())
218 Adding metadata to a buffer can be done with the gst_buffer_add_meta() call.
219 This function will create new metadata based on the implementation specified by
220 the GstMetaInfo. It is also possible to pass a generic pointer to the add_meta()
221 function that can contain parameters to initialize the new metadata fields.
223 Retrieving the metadata on a buffer can be done with the
224 gst_buffer_meta_get() method. This function retrieves an existing metadata
225 conforming to the API specified in the given info. When no such metadata exists,
226 the function will return NULL.
228 GstMetaTiming *timing;
230 timing = gst_buffer_get_meta (buffer, GST_META_TIMING_INFO);
232 Once a reference to the info has been obtained, the associated metadata can be
233 added or modified on a buffer.
235 timing->timestamp = 0;
236 timing->duration = 20 * GST_MSECOND;
238 Other convenience macros can be made to simplify the above code:
240 #define gst_buffer_get_meta_timing(b) \
241 ((GstMetaTiming *) gst_buffer_get_meta ((b), GST_META_TIMING_INFO)
243 This makes the code look like this:
245 GstMetaTiming *timing;
247 timing = gst_buffer_get_meta_timing (buffer);
248 timing->timestamp = 0;
249 timing->duration = 20 * GST_MSECOND;
251 To iterate the different metainfo structures, one can use the
252 gst_buffer_meta_get_next() methods.
254 GstMeta *current = NULL;
256 /* passing NULL gives the first entry */
257 current = gst_buffer_meta_get_next (buffer, current);
259 /* passing a GstMeta returns the next */
260 current = gst_buffer_meta_get_next (buffer, current);
268 We initially allocate a reasonable sized GstBuffer structure (say 512 bytes).
270 Since the complete buffer structure, including a large area for metadata, is
271 allocated in one go, we can reduce the number of memory allocations while still
272 providing dynamic metadata.
274 When adding metadata, we need to call the init function of the associated
275 metadata info structure. Since adding the metadata requires the caller to pass
276 a handle to the info, this operation does not require table lookups.
278 Per-metadata memory initialisation is needed because not all metadata is
279 initialized in the same way. We need to, for example, set the timestamps to
280 NONE in the MetaTiming structures.
282 The init/free functions can also be used to implement refcounting for a metadata
283 structure. This can be useful when a structure is shared between buffers.
285 When the free_size of the GstBuffer is exhausted, we will allocate new memory
286 for each newly added Meta and use the next pointers to point to this. It
287 is expected that this does not occur often and we might be able to optimize
288 this transparently in the future.
292 When a GstBuffer is freed, we potentially might have to call a custom free
293 function on the metadata info. In the case of the Memory metadata, we need to
294 call the associated free function to free the memory.
296 When freeing a GstBuffer, the custom buffer free function will iterate all of
297 the metadata in the buffer and call the associated free functions in the
298 MetaInfo associated with the entries. Usually, this function will be NULL.
304 When buffer should be sent over the wire or be serialized in GDP, we need a way
305 to perform custom serialization and deserialization on the metadata.
307 for this we can use the GValue transform functions.
313 After certain transformations, the metadata on a buffer might not be relevant
316 Consider, for example, metadata that lists certain regions of interest
317 on the video data. If the video is scaled or rotated, the coordinates might not
318 make sense anymore. A transform element should be able to adjust or remove the
319 associated metadata when it becomes invalid.
321 We can make the transform element aware of the metadata so that it can adjust or
322 remove in an intelligent way. Since we allow arbitrary metadata, we can't do
323 this for all metadata and thus we need some other way.
325 One proposition is to tag the metadata type with keywords that specify what it
326 functionally refers too. We could, for example, tag the metadata for the regions
327 of interest with a tag that notes that the metadata refers to absolute pixel
328 positions. A transform could then know that the metadata is not valid anymore
329 when the position of the pixels changed (due to rotation, flipping, scaling and
336 Subbuffers are implemented with a generic copy. Parameters to the copy
337 are the offset and size. This allows each metadata structure to implement the
338 actions needed to update the metadata of the subbuffer.
340 It might not make sense for some metadata to work with subbuffers. For example
341 when we take a subbuffer of a buffer with a video frame, the GstMetaVideo
342 simply becomes invalid and is removed from the new subbuffer.
345 Relationship with GstCaps
346 ~~~~~~~~~~~~~~~~~~~~~~~~~
348 The difference between GstCaps, used in negotiation, and the metadata is not
351 We would like to think of the GstCaps containing the information needed to
352 functionally negotiate the format between two elements. The Metadata should then
353 only contain variables that can change between each buffer.
355 For example, for video we would have width/height/framerate in the caps but then
356 have the more technical details, such as stride, data pointers, pan/crop/zoom
359 A scheme like this would still allow us to functionally specify the desired
360 video resolution while the implementation details would be inside the metadata.
366 We need to make sure that elements exchange metadata that they both understand,
367 This is particulary important when the metadata describes the data layout in
368 memory (such as strides).
370 We would like to use the bufferpool negotiation system to negotiate the possible
371 metadata that can be exchanged between elements.
373 When deciding the allocation properties, we will also negotiate the buffer
374 metadata structures that we can exchange.
380 Some structures that we need to be able to add to buffers.
383 * Arbitrary Matrix Transform
388 Some of these overlap, we need to find a minimal set of metadata structures that
389 allows us to define all use cases.