International organisation for standardisation organisation internationale de normalisation

27.2Contributions

Number	Session	Title	Source	Disposition

27.3Summary of discussions

27.4Action Points

28.Timed metadata metrics of Media in ISOBMFF (23001-10)

28.1Topics

28.1.1ISO/IEC 23001-10 Timed Metadata Metrics of Media in the ISO Base Media File Format

Specifies a storage format for commonly used, timed metadata metrics of media, such as quality related PSNR, SSIM, and others, for carriage in metadata tracks of the ISO Base Media File Format.

28.2Contributions

Number	Session	Title	Source	Disposition
m33185	File Format	Proposal of improvements to CD 23001-10	X. Ducloux P. Gendron J. Le Feuvre C. Concolato	noted
m33239	File Format	FF: Temporal Quality Signalling in ISO Based Media File Format	[e-mail], Zhu Li [e-mail], Imed Bouazizi	Accepted w14342
m33255	File Format	Improvements on CD text of Carriage of Timed Metadata of Media in ISOBMFF	[e-mail], Shaobo Zhang [e-mail], Xin Wang	Accepted w14342

28.3Summary of discussions

28.3.1m33255 Improvements on CD text of Carriage of Timed Metadata of Media in ISOBMFF

We are concerned that the MOS score doesn’t say what the panel size was, or alternatively if it met the requirements of some standard. We note that each metric would need revision to say how the accuracy applies, and what restrictions on the range of accuracy values there might be, and so on. The revised sample format also needs its pseudo-code, and we are concerned that adding a 3-byte accuracy value to each (typically 1-byte) metric value enlarges each (admittedly tiny) sample.

In terms of file format, we could make this work (with adjustments to the text), but we’d like the DASH group to comment on the general applicability of accuracy measures.

Thank you for the typo report: into the study.

28.3.2m33185 Proposal of improvements to CD 23001-10

There may be a bug in part 12: metadatasamplentry might not explicitly say it takes extension boxes, which we use here. Agreed, we don’t need an array for the green metadata. Each green metric has its own sample entry code, and its own sample format. Agreed, it would be good to use ‘configurationbox’ as part of the names of such boxes. We need a revision into a proposed study text (later in the week).

28.3.3m33239 FF: Temporal Quality Signalling in ISO Based Media File Format

We found this a little confusing as the use case in a file format – where frame loss doesn’t happen as such – is not clear. It’s more bitstream thinning or pruning, perhaps. Perhaps in DASH one would build subset representations, discarding the highest tiers successively down to the lowest. Do we need temporal quality information for that (and we note it would be relayed to the client, even though it’s really only used to subset representations)? This might help the server build the representations, preserving some ‘technically discardable’ frames that are ‘quality important’. We’d like the complete use-case assembled in the DASH group to make sure that this is the right information.

28.4Action Points

ISO/IEC CD 23001-10 (SC 29 N 14045)

Part 10: Carriage of Timed Metadata Metrics of Media in ISO Base Media File Format

CD (2014-05-05)