Experts are urged to study these issues (see mandate). Also worth considering late binding, and files that are delivered multiple ways. Do we need pdin at the fragment level? Do we need movie-fragment relative addressing where the moof is after the media data rather than before it? See also the defect report, as some of that introductory material proposed could cover some of these issues.
m35247 Summary of Voting on ISO/IEC 14496-12:2012/DCOR 3 and ISO/IEC 15444-12:2012/DCOR 3
Disposition: accepted
All approved, thank you.
m35087 Clarifications on ISOBMFF negative composition offsets
It seems that there is some confusion over whether the cslg box is descriptive or prescriptive. The only equation in the spec says CTS=DTS + CTSoffset, but this is by no means clear. Also the CTTS box has a text error (it promises that the offsets are unsigned). We need to document that the cslg is informative, and (for example), you could subtract the shift from the DTS values if you want DTS timestamps and you want them all before or at CTS. The ‘model’ that we expect many systems to use decode ordering (but not timestamps) and only the CTS timestamp is nowhere stated! We write a Defect Report document.
m35138 Comment on Edit List in ISO Base Media File Format
Yes, we seem to have lost the text that a movie fragment extends the last edit in the edit list. We have never worked out how to document edits in movie fragments, by the way. We should delete the phrase “(could be the length of the whole track)”. We need a paragraph on the interaction of fragments and edit lists, making this all explicit. So, the edit list can only talk about media in the movie (if any), and then fragments “push it out”. So, if there is media in the initial movie and the last edit does not select to the end of it (e.g. 100 seconds of media, but the last edit selects only the first 50 seconds), does a subsequent fragment (a) get ignored, as the previous edit didn’t go to the end or (b) insert an implied edit that is the duration of the fragment. We need to make sure that the phrase ‘empty edit’ is not ambiguous between (i) an edit that has duration but selects no media (i.e. nothing plays for that time) and (ii) an edit that has zero duration.
In the examples, we need to use the field names correctly.
Into the Defect Report [Dave]. We think that (b) above is what we will write, and we ask experts to check whether this is in line with current expectations.