EP1576607A1

EP1576607A1 - Characteristic point information (cpi) for multilayer video

Info

Publication number: EP1576607A1
Application number: EP03813682A
Authority: EP
Inventors: Declan Patrick Kelly; Jan Friso R. Blacquiere
Original assignee: Koninklijke Philips Electronics NV; US Philips Corp
Current assignee: Koninklijke Philips NV; US Philips Corp
Priority date: 2002-12-19
Filing date: 2003-12-10
Publication date: 2005-09-21
Also published as: JP2006511026A; CN100472641C; KR20050120625A; AU2003303270A1; KR100987655B1; WO2004057613A1; CN1830033A; AU2003303270A8; WO2004057613A8

Abstract

A method for generating characteristic point information (CPI) for multilayer encoded audio/video data includes generating a CPI file. The CPI file includes at least a point type corresponding to at least one point in a clip, a presentation time corresponding to the point in the clip, and one or more offset points corresponding to the point in the base layer and one or more enhancement layers of the clip. The point identified by the offset in the base and enhancement layers of the clip all correspond to the same point in the clip.

Description

CHARACTERISTIC POINT INFORMATION (CPI) FOR MULTILAYER VIDEO

The invention relates generally to video encoding and more particularly to multilayer video encoding.

For portable storage, multilayer encoded video provides a number of advantages. In devices with small screens and low power, only the base layer is retrieved and displayed. For devices with bigger screens and more power, both the base layer and one or more enhancement layers can be retrieved and displayed. Although multilayer schemes are most effective for portable devices such as the small-form-factor optical discs ("SFFO"), it can also be used in fixed devices to achieve flexible storage schemes.

The following description considers in particular audio/video ("A/V") devices operating according to the MPEG standards (ISO/IEC 11172 for MPEG-1, and ISO/IEC 13818 for MPEG-2) although the skilled persons in this technology will recognize the applicability to other A V coding schemes not in conformance with the MPEG standard. A clip refers to all or a portion of an audio/video (A/V) file or stream, etc., which may be multiplexed. (A clip file may be more generally characterized as a content file.) The multiplexed stream of which it is a portion is compliant to its defining specification in all respects, for example, in the case of an MPEG-2 multiplexed stream it complies with the MPEG-2 Program Stream or Transport Stream specification. Decoding of some elementary streams in the multiplex may need to start at a later address in the clip because partial access units may be present at the start and end of a clip. Presentation may need to start at an even later presentation unit.

Each clip is stored in its own real-time file. The physical allocation of the real-time file on disc ensures a continuous supply of data for reading and writing. Parts of the data in a clip may be shared between two or more clips by using data sharing mechanisms defined in the file system. In general, an initial recording is made into a single clip, with the number of clips rising as edit operations are made.

Characteristic Point Information (CPI) generally refers to information that is used to represent characteristic points in clips or information files. Characteristic points correspond to locations of particular data elements in clip files, for example, points considered of interest for a particular reason. For example, for clips that contain video stream data, characteristic points are used to indicate the start or the end of certain video access units. CPI is generally used to determine the location of relevant data elements in a clip, without having to read and parse the clip itself. Thus, CPI is important for several different system operations.

Some examples of such operations include trickplay operation, interactive playback, and quick edit-like operations. During a trickplay operation, such as fast forward and fast reverse play of a clip that contains video stream data, the stream is not read and decoded completely. Instead, only selected parts of the stream are read and decoded. The selection of these parts is based on CPI. For this purpose, CPI includes, for example, characteristic point information that indicate the location of picture data.

In the case of interactive playback, the user may require playback to start at some particular time position in a clip, rather than at the beginning of the clip. In such a case, CPI is used to find the data locations for reading and decoding a location in the clip that match this time position. For this purpose, CPI may include, for example, time-stamp values for relevant audio and video access units.

An important characteristic of quick edit-like operations involving clip truncation is that they produce truncated versions of clips without actually touching any data in the clips themselves. Typically, these kinds of operations manipulate clip references to produce new clips with references to truncated versions of such clips. Possible examples are division of clips into smaller ones, creation of new clips by means of non-seamless edit connections, etc. In all of these cases, CPI may be used to determine where truncated versions of the clip should start and/or end. The specification of characteristic points for a stream generally depend on the type and the contents of the stream.

In video recoding systems such as HDD, and in a non-multilayer context, CPI is extracted from the medium and stored (for example, in RAM) to provide a mapping between the presentation time of the content and the location of the content on disc. CPI is stored in a file separate to the actual audio/video ("A/V") content. Typically CPI is used to point to I-fiames in MPEG-2, but in general CPI is used to point to interesting points in the content, for example, entry point or other key points. In existing systems without multilayer video, an entry in the CPI file typically defines point type such as an entry point ("I-frame"), presentation time ("PTS"), and offset within the A V file or absolute address for the point in the A/V file.

A conventional multilayer encoded video includes a base layer and one or more enhancement layers. Typically, the base layer is independently coded and the data encoded in the enhancement layers supplement the base layer data. Thus, the base layer may be separately decoded and utilized. However, in order to use an enhancement layer, the base layer and the enhancement layer must be decoded and utilized together. In general, a smaller or older device may use only the base layer in playing the clip, whereas larger and/or newer devices (having advanced processing and/or resolution) will use the base layer and one or more enhancement layers when playing the clip, with a commensurate improvement in quality. The number of enhancement layers utilized depends on the device used in playback.

There is a need to have CPI for both the low quality and higher quality encoding of a multilayer scheme. Accordingly, there is a need for an extended CPI scheme to take account of multilayer video.

Accordingly, a method for generating characteristic point information (CPI) for multilayer encoded audio/video data is provided. In one aspect, the method includes generating a CPI file. The CPI file includes at least a point type corresponding to at least one point in a clip, a presentation time corresponding to the point in the clip, and one or more offset points corresponding to the point in the base layer and one or more enhancement layers of the clip. The point identified by the offset in the base and enhancement layers of the clip all correspond to the same point in the clip (most readily conceptualized as the same presentation time). Alternatively, an actual presentation time (or an absolute memory location) corresponding to the point in the base and one or more enhancement layers of the clip may also be included.

The method in another aspect includes generating separate CPI files for the base layer and each of the enhancement layers. The CPI file for the base layer includes information about the points in the base layer, and likewise, the CPI files for the enhancement layers includes information about the points in the corresponding enhancement layers. The CPI files for the enhancement layers may include only the Offset information and may use the point type and presentation time for the corresponding point in the base layer CPI file.

Yet in another aspect, the method includes generating a CPI file to include information about a base layer, and generating separate additional CPI files to include information about each enhancement layer. The CPI file corresponding to a particular enhancement layer may, for example, include pointers to the enhancement layer and all lower layers, as well as the base layer.

These and other features of the present invention will be apparent to those skilled in the art from the following detailed description, taken together with the accompanying drawings, in which like reference numerals refer to like parts.

Figure 1 is a diagram illustrating a CPI scheme in one embodiment.

Figure 2 illustrates the CPI files in one embodiment corresponding to each layer.

Figure 3 illustrates a CPI structure having a separate CPI file for all enhancement layers in one embodiment.

Figure 1 is a diagram illustrating a CPI scheme in one embodiment. Briefly, for each clip or A/V file, there is an accompanying CPI file containing the information for the characteristic points in the clip file. A characteristic point (or "point") in the clip file has an entry in the CPI table structure that includes at least a point type (or if only one point type possible then it is implicit), a PTS and an offset within the file.

Accordingly, for each characteristic point, several parameters are used. Thus, each entry 104 in a CPI file 102 may include a point type, PTS, and one or more offsets. The point type may indicate, for example, that the point corresponds to the start position of an I frame. The points in the point type may also be in pairs, for example, start and end points of an I-frame. A single point may contain both a start and end pointer, or the same may be considered two points. If all characteristic points include both start and end pointers, then it is efficient to combine the start and end pointers for each point as a single point. If not all points include start and end pointers, then use of start and end pointers by certain points is acceptable.

PTS is a data parameter used in an entry in the CPI. The PTS in this case represents the presentation time well known in MPEG. For example, PTS may be a 32-bit time-stamp value representing the presentation time of the I-frame. This parameter generally represents the data field that is associated with the point. The exact syntax and semantics of the data field depend on the type of the point and is specified for each type separately. Thus, the point type and point data parameters correspond to one another. In addition, the offsets included in the entries of the CPI file correspond to a location in a respective video layer. In particular, Offset B identifies the location corresponding to the point of entry 104 in the Base layer 106, Offset 1 identifies the location of the point of entry 104 in Enhancement layer 1 108a, etc. The offset value included in the entry for each layer gives the offset within the layer's file corresponding to the point of the entry. As is known in the art, the offset defines how far forward to move within the file starting at a known starting point. In this case, starting at the known starting point and moving forward by the offset in the layer file results in moving to the point under consideration in the entry. The offset is typically used in an application and subsequently translated to an absolute address when accessing the data in the storage medium (such as a disc). Thus, the offsets included in each entry provide the respective locations in the files for the base and enhancement layers that correspond to the characteristic point of the entry. Alternatively, the absolute address of the corresponding location for each layer may be used in the entry instead of the offset. This allows the base and any enhancement layers for a point to be accessed using the entry for the point in the CPI file. To free space on the disc, a user may delete a highest level enhancement layer, for example, 108n in Fig. 1. In this case, some of the offsets within the CPI file may no longer be valid because they may refer to a deleted file. Thus, the CPI file may also include an indication of the layers that are valid (represented by the presence of a check mark for valid layers in the entry of Fig. 1). Since the number of layers in the CPI file depends on the number of layers present when recording, which may vary per recording, the CPI file structure also may include an indication of the number of layers referred to.

Entries 104 of the CPI file 102 of Fig. 1 may further be efficiently configured in a horizontal manner. For example, the PTS of each entry may include an absolute time value. Alternatively, the PTS for every nth entry may be an absolute value, and the PTS for subsequent entries in the CPI file may be relative to the last absolute value (referred to as "horizontal offset"). For example, the PTS may be an absolute value every 10th entry. For the nine entries in the CPI file following the last absolute PTS entry, the PTS may be given as a horizontal offset with respect to the last absolute PTS. Other parameters in the entries of the CPI file may be configured in like manner.

For example, Offset B may be an absolute offset value in the CPI file every 10th entry. For the nine entries in the CPI file following the last absolute offset value for Offset B, Offset B may be given as a horizontal offset with respect to the absolute Offset B. For example, an entry n in CPI file 102 may include an absolute value for Offset B of 1000 and entry n+10 includes the next absolute value for Offset B of 1100. The intervening entries n+1, n+2, ..., n+9 have horizontal offset values for Offset B of 10, 20, ..., 90, instead of absolute values for Offset B of 1110, 1120, ..., 1190. Thus, when the application reads the parameter for Offset B of entry n+2, for example, it retrieves the horizontal offset 20 from the entry and determines an Offset B of 1120 based on the absolute offset value for Offset B in entry n. Offset B of 1120 is then used as the offset to access the corresponding location in Base layer 106 for entry n+2.

Such a horizontal configuration between entries in the CPI file may likewise be used for the other parameters and further reduces the required memory without a significant decrease in performance. It is noted, however, that because the base and enhancement layer files have different lengths, the horizontal offset between CPI entries for one parameter cannot be used for another parameter. For example, the horizontal offset for Offset B for entry n+2 (20 in the above example) generally cannot be used as the horizontal offset for Offset 1 for entry n+2, since the Base and first enhancement layer files will have different lengths.

Returning to the focus of "offset" as shown in Fig. 1, namely as pointing to a location in the base and enhancement layers corresponding to a point of an entry, the CPI file structure of Fig. 1 is generic and may be applied to other existing CPI schemes, for example, those using naming schemes. For example, in certain systems the association between the CPI and the A/V content is based on file names, or more particularly, file name extensions. For example, file.clpi may denote file clip information and file.m2ts may denote file MPEG-2 transport stream. For multilayer video, this scheme may be enhanced, for example, as including file.clpi for file clip info, file.av for file base layer A/V data, filel .av for file enhancement 1 A/V data, and file2.av for file enhancement 2 A/V data, etc. In such a multilayer application, file.clpi would be modified to include layer offsets for each entry, as shown in Fig. 1. Thus, each characteristic point entry in file.clpi would include a base layer offset (Offset B) corresponding to file.av, a first enhancement layer offset (Offset 1) corresponding to file av, etc. In addition, the above description assumes that each enhancement layer has a characteristic point at the same time location, which are all stored in the CPI entry. Some of the offset parameters may be omitted for some entries for higher level layers, in order to reduce the size of the CPI file. For example, the offset for the first enhancement layer may only be included in every second or third entry. This, however, will restrict the characteristic points that can be used in certain functions, such as high quality trickplay. In another aspect, the CPI may be stored for each layer separately on disc, so for each file of layered encoding, there will be a corresponding CPI file. Figure 2 illustrates the CPI files in one embodiment corresponding to each layer. The base file 206 has a corresponding CPI file (CPI FileB 202), separate from the enhancement layers 208a, 208b. The first enhancement layer Enh 1 file 208a has its own CPI file (CPI Filel 204), as does the Nth enhancement layer Enh N 208b file whose corresponding CPI file is shown as CPI FileN 210.

There are a number of advantages of using separate CPI files for each file of layered encoding. Enhancement files can be deleted without the need to note the deletion in the CPI files for other layers. In addition, there is no need to read the CPI file for enhancement layers that are not utilized. Having a separate CPI file corresponding to each layer eliminates potential problems and or adjustment of the CPI file when one or more enhancements are deleted. For example, when one or more enhancements are deleted, a composite CPI file such as shown in Fig. 1 would then include one or more invalid parameters in the entries for the now deleted enhancement layers. (As noted above, an indication of invalid layers would then be included in the CPI.) Also, if all CPI are contained in one CPI file, it may be necessary to read the complete CPI file even when only the base layer is being played. In one aspect, the similar file naming conventions described in the above example may be used. For example, file.clpi may be used as the CPI file for the base file (202 Figure 2), filel.clpi for file enhancement 1 clip info (204 Figure 2), fileN.clpi for file enhancement N clip info (210 Figure 2), file.av for file base layer A/V data (206 Figure 2), filel .av for file enhancement 1 A/V data (208a Figure 2), and fileN.av for file enhancement N A/V data (208b Figure 2).

In one case for the embodiment of Fig. 2, the point type and PTS are duplicated for each CPI file 202, 204, 210. (Each CPI file also includes the offset value for the respective layer.) In another aspect, duplicating the point type and PTS in each CPI file may be avoided. Thus, the base layer CPI 202 may include point type, PTS, Offset B (for the base layer), and for each enhancement layer, the corresponding CPI file may have entries that include the offset for the layer. In this case, it is understood that an entry n in the base CPI file corresponds to entry n in each of the enhancement CPI files. In addition, the point type and PTS for entry n in the base CPI file applies to entry n in each of the enhancement CPI files. In this way, the point type and PTS need not be duplicated for each CPI file (namely, for CPI Filel - CPI FileN).

Even though all parameters for a point are not included in a single entry in a single CPI file (as in the embodiment of Fig. 1), once the separate CPI files are loaded in memory, there is no performance penalty for utilizing a separate CPI structure, as shown, for example, in Fig. 2. Corresponding entries in the various CPI files may be readily accessed for the same point once loaded, where the loading of the separate CPI files in memory comprises a sequential combination of the CPI files. For example, if there are M CPI entries, the base layer CPI has an entry size A for characteristic point information and the first enhancement layer CPI has an entry size B for characteristic point information, then to access to an entry for a particular point x from the combined CPIs starting at address S in memory, the following operation may be performed: x.Point Type = (S + A*x).Point Type x.PTS = (S + A*x).PTS x.Base Offset = (S + A*x).Base Offset x.Enh Offset = (S + A*M + B*x).Enh Offset

Similarly, the entries of CPI files for multiple enhancement layers may be accessed with analogous simple calculations. Thus, performance is not sacrificed by the storing of separate CPI files as described above on the disc (or other medium). In addition, the horizontal offsets used between CPI entries described above applies equally well in the case of separate CPI files for each layer.

As noted, the above operation assumes that the CPI files are loaded in memory so that the enhancement CPI files follow directly from the Base CPI files. If they do not follow, however, a new offset denoting the memory distance between the files may be included in the calculations.

It is implicit in the above structure that the number of entries in the base and enhancement layers are both equal to M. However, an enhancement layer may include a lesser number of characteristic points, for example, one characteristic point for every 2 or 3 points in the base layer (corresponding to the frequency of sub-sampling). In that case, if the number of CPI entries for enhancement layer is M, then every two or three points in the enhancement layer may be the same. This is an unnecessary use of memory. However, it is still possible with the above structure to have less entries for the enhancement layer (by eliminating the adjacent redundant 2 or 3 points in the enhancement layer) than the base layer as long as the frequency of the sub-sampling is known and taken into account in a modification of the above operation.

In another aspect, a CPI file may include all lower layer points for the enhancement layers. This means that each CPI file is self contained and includes pointers to the current layer and all lower layers. This gives the possibility to include more entries for characteristic points in the CPI for the enhancement layers than the base layer, as described further below. Figure 3 illustrates this scheme in one embodiment. A CPI file 302 is used for a base file 306. An entry 304 in the CPI file 302 includes information about a point in the base file 306. A separate CPI file 310 is used when enhancement layer 1 308a is utilized. An entry 312 in this CPI file 310 includes information about both base layer 306 and Enh 1 308a. Similarly inclusive CPI files would exist for higher level enhancement layers.

In general, to decode the enhancement layer, the corresponding base layer data is needed since extra pointers to the enhancement layer without corresponding pointers to the base layer are not useful. With the embodiment shown in Figure 3, the base layer CPI 302 may be very small, for example, entries for a point only every few seconds. This may be sufficient for random access and low quality trickplay since the base layer is generally for use on small portable devices with tight power constraints. Also, having small CPI reduces the memory needed to store CPI entries. In such cases, CPI 302 that includes the offset for the base layer is retrieved and used. For the base layer plus enhancements, more points may be used to ensure high quality trickplay when played on devices with bigger screens and more power, thus resulting in a CPI file with more entries (as well as more parameters per entry). For example, when the device utilizes a first enhancement layer, CPI 310 is retrieved and utilized, thus providing the CPI data for both the base layer and first enhancement layer..

In the embodiment of Fig. 3, the content file corresponding to CPI 310 may alternatively be a composite encoding of the base layer 306 and 308a. Thus, separate content files may exist for the base layer, the base layer plus the first enhancement layer, the base layer plus the first and second enhancement layers, etc. File 306 is decoded when only the base layer is utilized. If the first enhancement layer is used, then the compositely encoded base and Enhl file is decoded and used. Each such content file may have a separate CPI file that includes the point type, PTS and offset for the particular base layer file or composite base and enhancement layer(s) file.

While the invention has been described with reference to several embodiments, it will be understood by those skilled in the art that the invention is not limited to the specific forms shown and described. Thus, various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

CLAIMS:

1. A recording medium including a number of content files including a base layer (106) and one or more enhancement layers (108a, 108n), the medium further including a characteristic point information (CPI) file (102) for the base layer (106) and one or more enhancement layers (108a, 108n).

2. The recording medium as in Claim 1, wherein the CPI file (102) is comprised of a sequence of entries (104) each corresponding to a characteristic point in the base layer (106) and the one or more enhancement layers (108a, 108n), the entry (104) for each characteristic point comprised of a point type, a presentation time (PTS) and a separate offset for each of the base layer (106) and at least one enhancement layer (108a, 108n).

3. The recording medium as in Claim 2, wherein, for each entry (104), the offset for the base layer (106) determines a memory location in the base layer (106) for the characteristic point of the entry (104), and the offset for each enhancement layer (108a, 108n) determines a memory location in the respective enhancement layer (108a, 108n) for the characteristic point of the entry (104).

4. The recording medium as in Claim 3, wherein the memory location in the base layer (106) for the characteristic point of the entry (104) and the memory location in each enhancement layer (108a, 108n) for the characteristic point of the entry (104) correspond to the presentation time of the entry (104).

5. The recording medium as in Claim 3, wherein for at least a number of the entries (104) in the CPI file (102), the offset for at least one of the base layer (106) and the enhancement layers (108a, 108n) is an absolute memory address in the respective layer file for the characteristic point of the entry (104).

6. The recording medium as in Claim 3, wherein for at least a number of entries (104) in the CPI file (102), the offset for at least one of the base layer (106) and the enhancement layers (108a, 108n) is a measure from a known starting point within the respective layer file, the measure providing the memory address in the respective layer file for the characteristic point of the entry (104).

7. The recording medium as in Claim 1, wherein the CPI file (102) includes an indicium that one or more enhancement layers (108a, 108n) are deleted from the recording medium.

8. A recording medium including a number of content files including a base layer (206) and one or more enhancement layers (208a, 208b), the medium further including a separate characteristic point information (CPI) file (202, 204, 210) corresponding to each of the base layer (206) and one or more enhancement layers (208a, 208b).

9. The recording medium as in Claim 8, wherein the CPI file (202) corresponding to the base layer (206) is comprised of a sequence of entries each corresponding to a characteristic point in the base layer (206), the entry for each characteristic point comprised of a point type, a presentation time (PTS) and an offset for the base layer (206).

10. The recording medium as in Claim 9, wherein the CPI file (204, 210) corresponding to each of the one or more enhancement layers (208a, 208b) is comprised of a sequence of entries corresponding to a characteristic point in the respective enhancement layer (208a, 208b), the entry for each characteristic point comprised of an offset for the respective enhancement layer (208a, 208b).

11. The recording medium as in Claim 10, wherein each entry for a particular characteristic point in the CPI file corresponding to each of the one or more enhancement layers (208a, 208b) corresponds to an entry for the particular characteristic point in the CPI file corresponding to the base layer (206).

12. A recording medium having two content files including a base layer (306) and a first enhancement layer (308a), the medium further including two characteristic point information (CPI) files (302, 310), the two CPI files (302, 310) comprised of a base CPI file (302) corresponding to the base layer (306) and a first CPI file (310) corresponding to the base layer (306) and first enhancement layer (308a).

13. The recording medium as in Claim 12, wherein the base CPI file (302) is comprised of a sequence of entries each corresponding to a characteristic point in the base layer (306), the entry for each characteristic point comprised of a point type, a presentation time (PTS) and an offset for the base layer (306).

14. The recording medium as in Claim 12, wherein the first CPI file (310) is comprised of a sequence of entries each corresponding to a characteristic point in the base layer (306) and the first enhancement layer (308a), the entry for each characteristic point comprised of a point type, a presentation time (PTS) and a separate offset for each of the base layer (306) and first enhancement layer (308a).

15. The recording medium as in Claim 12, wherein the recording medium includes one or more higher level enhancement layers, the medium further including an additional CPI file corresponding to each of the one or more higher level enhancement layers, the CPI file for each higher level enhancement layer corresponding to the base layer (306) and all enhancement layer levels (308a...) up to and including the corresponding higher level enhancement layer.

16. A method for recording multilayer encoded audio/video data, comprising: generating a CPI file (102), the CPI file (102) including a sequence of entries (104) each corresponding to a characteristic point, each entry (104) comprised of a point type, a presentation time (PTS), and one or more offsets corresponding to the characteristic point in a base layer (106) and one or more enhancement layers (108a, 108n).

17. A method for recording multilayer encoded audio/video data, comprising: generating a base CPI file (202) corresponding to a base layer file (206) and including a sequence of entries each corresponding to a characteristic point, each entry comprised of a point type, a presentation time (PTS) and an offset corresponding to the characteristic point in the base layer (206); and generating one or more additional CPI files (204, 210) corresponding to a respective one or more enhancement layer files (208a, 208b) containing enhancement layer audio/video data, each additional CPI file (204, 210) corresponding to an enhancement layer file (208a, 208b) comprised of a sequence of entries including at least an offset corresponding to a location in the corresponding enhancement layer audio/video data.

18. The method of claim 17, wherein each entry in each additional CPI file (204, 210) for an enhancement layer file (208a, 208b) corresponds to an entry in the base CPI file (202) for a characteristic point.

19. The method of Claim 18, further including the steps of: accessing an entry for a characteristic point in the base layer CPI file (202) and using the offset in the entry to access the location corresponding to the characteristic point in the base layer file (206), and accessing a corresponding entry for the characteristic point in at least one additional CPI file (204, 210) for an enhancement layer (208a, 208b) and using the offset in the corresponding entry to access the location corresponding to the characteristic point in the corresponding enhancement layer (208a, 208b).

20. The method of Claim 17, further including the steps of: storing the base CPI file (202) and the one or more additional CPI files (204, 210) in sequence in a memory, and applying an operation based on the stored sequence to access an entry in the base CPI file (202) and corresponding entries in the one or more additional CPI files (204, 210).

21. A method for recording multilayer encoded audio/video data, comprising: generating a base CPI file (302) corresponding to a base layer file (306) and including a sequence of entries each corresponding to a characteristic point, each entry comprised of a point type, a presentation time (PTS) and an offset corresponding to the characteristic point in the base layer (306); and generating one or more additional CPI files (310) corresponding to a respective one or more enhancement layer files (308a) containing enhancement layer audio/video data, each additional CPI file (310) corresponding to an enhancement layer file (308a) comprised of a sequence of entries each corresponding to a characteristic point, each entry including a point type, a PTS, and offsets corresponding to the location of the characteristic point in the base layer (306) and all enhancement layer files (308a) of a level up to and including the corresponding enhancement layer file.

22. An information structure for identifying multilayer encoded audio/video data, comprising: a CPI (102, 204, 210, 310) including at least a point type corresponding to a point on a clip, a presentation time corresponding to a point in the clip, and one or more offset points corresponding to a point in one or more enhancement layers (108a, 108n, 208a, 208b, 308a) of the clip.