Open TanguyPerson opened 5 years ago
Besides, it is still unclear to me why this explicit main Role signaling is required for video... and why it is not required for audio !
A typical use case would be where same audio is available in 2 codecs (e.g. AAC and E-AC-3). It requires using 2 separate Adaptation Sets (due to codec). If I follow the logic described for Video, I should not rely on default Role being 'main', and add explicit 'main' Role descriptors, to remove any ambiguity ?
Note: my typical use case for video is when having different video Representations (SD/HD/UHD) scrambled in CENC with different keys, which requires using one video Adaptation Set per key.
I believe the intent of the spec is to say that default is "main" and you don't have to signal "main".
In your described scenario, I would expect that you do not signal any role for any video adaptation set - they are all main content, which is the default role.
Agreed that the wording is quite confusing. I would clarify wording by:
For video, default video Role is defined as 'main' in some parts of specifications [1] [2]. But other parts [3] [4] deny that default Role value and require explicit Role descriptor for video.
This is at least confusing. IMHO, is is even contradictory : the default value is not always applied !
Could we improve this ? Proposal for § 3.9.2 :
Any better wording is welcome.
[1] Table 24 in § 3.9.2:
[2] §3.9.4.2
[3] § 3.2.1
[4] §3.2.2.1