International organisation for standardisation organisation internationale de normalisation

12.3Summary of discussions

m18464: the flag will be considered as part of Representation ‘annotation” in more generalized case and will be discussed there.
m18465: will bring back an enhanced version of proposal that also includes less on MPD signaling.

Show the evidence of how often the client can create a valid MP4 file, using the existing players (whether file with missing track in mdat can be played by legacy player?)

-“If satisfied, proposal 1 (one initialization segment for all reps) get added as client/service guidelines (informative)” no input.

ExpriationTimeUTC: Samsung will have contribution at the Jan meeting.

m18353: seems not directly related to EE8
m18562 moved to EE1 (CMP).
m18370: supporting information + index file+ MPD additions for index file

m18462: dual approach using sidx and MP4 or separate file

m18457: two solutions: 1) index file similar to m18370 2) include RAP index in PID stream, prefer PID approach. Need to transcode new TS.

:m18558 & m18560: fixes and improvements on the baseline.

m18392: extensions 2 M2TS, goes to M2TS extension work, outside of DASH
m18559:similar to m18392 but use private data
m18545: the audio glitch during switching of video streams
m18566: - map PTS to MPD timeline, gzip vs LG proposal: not completed.
m18439: ETRI, proposes “periodref”, Xlink and XInclude are not appropriate.

m18551: section 3. Propose a new element “Include” in MPD inside any period or representation.

Both 18439 and 18551 do not recommend XInclude or Xlink.

Some people think Xlink may able to do same. Will bring evidence tomorrow.

Will decide on the solution after CMP review.
M18388: timemetadata, no consensus yet.
m18366: msra box for accessing RAP. No major advantage to WD is shown yet.

m18387: new proposal to extend sidx box, will be reviewed with TRA proposals.

m18440: no strong support

m18565: no new decision

m18385: mediacomposition group

m18444: DID, MPD selection

m18498: aspect ratio. New attributes for signaling the aspect ratio.

m18551: RepresenationGroup

The concept of RepresentationGroup in m18551 was originally introduced by ETRI in section 5.4 of m18009 as componentRepresentations

Following attributes in RepresentatonGroup element (m18551) was originally introduced by ETRI in section 5.4 of m18009;

lang

mimeType

minBandwidth

maxBandwidth

minWidth

maxWidth

minHeight

maxHeight

minFramerate

maxFrame
m18562: view priority and PIP location are provided as representation attributes in MPD. But since MPD is not designed to convey presentation/layout information, therefore the proposed attributes are not added to MPD.

m18385 will provide a use case and detail example of its proposal compared with WD by Tues.
DRM part of CMP will be reviewed during the week.
18390: adding RAPperiod to MPD. Accepted the maxmiumRAPperiod into MPD.

Keep the constraint of one minimum RAP.

18401: minimum RAP frequency.

m18347: rewind trick mode: alternatePlayoutRate have negative value + additional text to clarify the meaning of negative/positive. MP4FF needs also to signal for reverse sample order in reverse order.

m18441: remove the constraint of having minimum one RAP. Propose RandAcess element in MPD to signal segments with RAPs.

m18443: temporal ordering of tick mode. Not to use one trun. Sample grouping using “tele” box in MP4. Use segment type bx to brand the segment as temporal scalability. A new attribute in trick mode element of MPD for temporal ordering, also number of levels.

Sample grouping functionality might exist in AVC FF. Accept m18443 with the refinement on AVC extension box, resolve the disagreement on index box after sidx. Collaboration with 18387.
Update on m18016:

m18387: extend the sidx box.

Decision:
m18441: bring back the proposal with more details.
A basic question to answer generally: if the values of attributes are out of range, what would be the behavior of the client? (contributions are welcomed to answer this).
m18463: need two decoded 2 I frames sometimes to switch seamlessly. To avoid overlapping, proposing bitstreamStrcutureId attribute for representation meaning that they can be switched and the old reference frame used for new ones.

m18493: sidx is not easy to create. MPD gives the accurate time for segments. Add the tfdt box after Track Fragment Header box. Tfad express the clean start.

m18494: switching while overlapping representation. encodedSampleNooverlapFlag at represenationgroup
m18496: sidx box has two purposes: timing recovery and RAP indexing. TrackFragmentDecodeTime can do it better.

Sidx Define decoding time, should have defined the composition time.

Accept it to WD. Ask 3gpp to align the 3GPP release 9.
Come back with fragmentAlignmentFlag to propose.

m18497: for information. Will be presented later.

M18405

Multiple URl with simple priority.

M18495

. If there is any issues with CDN, it will brought up. We will look at the exact syntax and accept it if there is no issue.

M18497

Add to WD: add “duration” to period.

Make period start time optional

Only for VOD case, if it doesn’t break anything else in MPD.

M18445: - supp. Can be delivered as content, no need to be part of MPD.

- Average bandwidth: not average. If use the same definition of Rep’s bandwidth. No consensus on this. Bring modified proposal.

M18524: - no strong support for flag.

- Group: WD will be clarified that this is the case.

M18357: Add stereoVideo to represenation and generic FramePackingType

M18484: deferred to later.

M18549

Propose QoE metrics for informative part. Exact parameters to be further defined and refined.

Approve an EE on this: Mark to provide the description.

M18483

On, UsageRights, UsageRightUri

MPDIntegrity: Need further studies. Possibly start an EE.

M18385

On downloadableDRMInfo

M18015 was presented in Paris and reviewed here again. DASH Media and DASH Client imple,ments the major feature of the software. Built on Mac, Linux.

m18402: duplicate registration.

m18403: the software plan

m18442: Test sequences and a site that provide VOD and live content for DASH.

m18444: Demoed a player which parses MPD and plays on representation.

M18381: implementation based on GPAC open source project. Parsing MPD. It doesn’t support switching

- m18309:

• clalify spec that track fragment boxes covering the same time period: corrigenda issue

• restriction on the composition time of following samples

• Movie fragments covering the same time period should be time-aligned. -> this should be addressed in DASH spec.

• Allow gaps in movie fragment box: corrigenda issue.

No additional FF amendment required. Either corrigenda issues or better DASH.

M18348: signaling rap in FF. 1) Define RAP to include all 3 types of RAPs. 2) Sync sample table box should have only IDR RAPs. 3) Add 2 fields to rap sample group entry. 4) Add one filed to syntax of roll sample group entry. Discuss with dave. May be already addressed. 5) Clarify raps in mfra ox can be any raps.
Give a definition of sync sample. Part of corrigenda. Editors need to go look at Part 12 and 15 and see what is missing about RAPs and propose for next corrigenda.
m18391: 3 new boxes for 3-D signaling. Svmi, scdi and mfsi

see part 15 extension and if it can be addressed there.

m18018 and m18019:
-Trik box: start work with RAP sample group box. Integrated sample group to be defined to address this. Will be addressed as part of editing of the PDAM.

-sample encryption box -> study if it can be implemented with time metadata track,

- track encryption box add to PDAM,

- pssh: draft a metada item, instread of using uuid, use a 4cc register id. Goes to PDAM.

m18418: HTTPHintSampleEntry, not part of delivery.
M180415: signaling the RAP indexes in MPD.
There are 3 different methods: inside M2TS signaling, sidx based signaling and signaling in MPD.