Re: WebRTC-SVC: "S" modes and simulcast using a single SSRC

Amit said:

"The RIDs aren't necessary in all packets of the same stream. in fact once the receiver acknowledges receipt of a stream with a MID or RID, the sender can remove the MIDs and RIDs from future packets carrying the same SSRC (as the relationship between SSRC -> MID and SSRC -> RID has been established by the receiver). to that extent, you should not send different layers using the same SSRC and relying on them having different RIDs, as receivers might also be ignoring the RID and MID extensions once the relationship to SSRC is established. Not sure if SVC would be different than regular simulcast in this regard, but worth mentioning."

[BA] Right.  I believe that both "S" and SVC modes are not reflected in SDP Offers and Answers. Or at least that's what we have been assuming in developing the examples of simulcast negotiation<https://github.com/AOMediaCodec/av1-rtp-spec/pull/60> within the AV1 RTP specification.
________________________________
From: Amit Hilbuch <amithi@google.com>
Sent: Tuesday, July 30, 2019 11:10 AM
To: Bernard Aboba <Bernard.Aboba@microsoft.com>
Cc: Florent Castelli <orphis@google.com>; Sergio Garcia Murillo <sergio.garcia.murillo@gmail.com>; WebRTC WG <public-webrtc@w3.org>
Subject: Re: WebRTC-SVC: "S" modes and simulcast using a single SSRC

The RIDs aren't necessary in all packets of the same stream. in fact once the receiver acknowledges receipt of a stream with a MID or RID, the sender can remove the MIDs and RIDs from future packets carrying the same SSRC (as the relationship between SSRC -> MID and SSRC -> RID has been established by the receiver).
to that extent, you should not send different layers using the same SSRC and relying on them having different RIDs, as receivers might also be ignoring the RID and MID extensions once the relationship to SSRC is established. Not sure if SVC would be different than regular simulcast in this regard, but worth mentioning.

On Wed, Jul 24, 2019 at 11:30 AM Bernard Aboba <Bernard.Aboba@microsoft.com<mailto:Bernard.Aboba@microsoft.com>> wrote:
Florent said:

"I don't think there are any issues adding S modes as it's just a special case of spatial layers which we intend to support already."

[BA] It sounds like there is interest in adding the "S" modes back.  I will prepare a PR for that and would appreciate your review.

"I would love to continue the discussion about this at TPAC and possibly include some time to discuss APIs to improve the application's control over each spatial or temporal layer's parameters."

[BA] If you have specific Issues or PRs relating to the WebRTC-SVC draft, can you file them in the repo?  TPAC is scheduled earlier this year so it will be upon us before we know it 😉