Valid targets for `cdsc` reference type

[bradh]

cdsc is used as the reference type for track reference and also for image items, plus things like regions.

Is it intended that cdsc should only be used where the target is a track or image-like item?

That is, it is valid for an item to describe another item that is not an image item? For example, metadata describing another metadata item?

[leo-barnes]

I think it's perfectly valid to have metadata describing metadata from a purely syntactical view. But none of the metadata types in HEIF/ISOBMFF really make sense to "nest", so it will probably only make sense to use with custom metadata payloads, which are non-portable in any case.

[bradh]

I note that some of the files in https://github.com/MPEGGroup/FileFormatConformance/pull/108/files have nested cdsc. For example, the region-polygon.heic example has Item 2 -[cdsc]-> Item 100, and Item 100 -[cdsc]-> Item 1. Item 1 and 2 are hvc1 and Item 100 is rgan.

[fmaze]

Thanks for this example, unfortunately, it seems this conformance vector is currently invalid. In region-polygon.heic example, we should rather have Item 100 -[eroi]-> Item 2, and Item 100 -[cdsc]-> Item 1.

In 23008-12 2nd ed. clause 6.10.3, (and future update in latest draft of 23008-12 2nd ed. FDAM1 clause 11.3.1), it is stated that region item is associated with:

• "metadata items, using an item reference of type 'cdsc' from the metadata item to the region item"
• "image items (...), using an item reference of type 'eroi' from the region item to the image item (...)"

[podborski]

Files were fixed in m67639 https://github.com/MPEGGroup/FileFormatConformance/pull/137

[cconcolato]

@bradh given that the files are now fixed, do you think the text needs update?

[bradh]

I'm OK with the updated wording.