**Session Date/Time:** 13 Feb 2024 17:00

# [AVTCORE](../wg/avtcore.html)

## Summary

The AVTCORE virtual interim meeting covered updates on several drafts, including v3c, Skip, J.2K, HEVC RTP for WebRTC, RTP over Quick, S-Frame, Green Metadata, RTCP Feedback Message Timing Configuration, and Haptics. Key discussions revolved around progressing drafts through the IETF process, resolving outstanding technical issues with ISG, and clarifying interactions with WebRTC and Quick. Several drafts are nearing Working Group Last Call or adoption.

## Key Discussion Points

*   **Administrative and Draft Status:**
    *   `bp9` is still in MISREF, awaiting frame marking.
    *   `EVC` has progressed to the RFC editor queue.
    *   `v3c` Working Group Last Call completed with sparse responses; more reviews from RTP experts are requested. InterDigital has a public implementation and demo, which they will link to the mailing list.
    *   Discussion on retiring the I-RTP payload format type registry was deferred to a non-interim meeting (IETF 119).
*   **Skip Update:**
    *   Version 8 was published, emphasizing the document's network purpose, reorganizing payload format, and clarifying optional use of congestion control, AVPF/SAPF, augmented transport protocols, and SRTP.
    *   Version 9 is in progress, addressing ISG feedback to soften security claims, focusing on payload encryption rather than general confidentiality.
    *   An ISG disclaimer text was discussed for inclusion.
    *   The ISG ballot has one 'yes' and seven 'no objection' positions; the remaining 'discuss' position needs to be cleared, with an aim to complete this before March ISG board changes.
    *   The design ensures that the inner media payload type is not encrypted hop-by-hop, allowing SFUs to read and rewrite it, effectively addressing a previously raised concern.
    *   Duplication of header extensions between inner and outer packets for SFU access was discussed, with a general inclination towards copying all for flexibility. This will be moved to the mailing list for further discussion.
*   **J.2K Updates:**
    *   Sample RTP packet streams, generated with a hardware implementation, are now available.
    *   An independent receiver implementation is a current focus.
    *   The chairs will set up a working group GitHub repository and investigate hosting large binary test vectors.
    *   More review from RTP experts is requested.
*   **HEVC RTP for WebRTC (JSEP Issues):**
    *   **Web Codecs Status:** Chrome supports hardware-only HEVC decode; Edge supports HEVC/AV1 decode. Safari Tech Preview 188+ includes H.265/AV1 encode/decode in hardware (H.265 lacks temporal scalability) but requires disabling "lockdown mode."
    *   **WebRTC Status:** Chrome has libwebrtc support for HEVC packetization (depacketization in review). Webkit has HEVC enabled by default but needs synchronization.
    *   **SPS/PPS/VPS Clarifications (Issue 20):** The draft recommends parameter sets (SPS/PPS/VPS) precede not only IDR frames but also CRA and BLA frames, due to potential parameter changes. A question was raised about whether open-GOPs are commonly used in web environments, potentially over-engineering this requirement.
    *   **Send-only/Receive-only Codecs and JSEP Interaction:** The challenge of asymmetric codec capabilities (e.g., Chrome decode-only HEVC, Safari encode/decode HEVC) and their interaction with JSEP's offer/answer model was discussed. The expected behavior for a symmetric answer (Safari sends H.265, Chrome sends H.264) was affirmed.
    *   The "setCodecPreferences" API's different implementations were noted as a source of issues.
    *   A general principle was articulated: protocol design should not be dictated by short-term implementation snapshots.
    *   The RTCWeb working group is moving to the WITS area and is not currently meeting, raising questions about where JSEP-related issues should be resolved.
*   **RTP over Quick:**
    *   Progress includes 12 open issues, 5 planned pull requests, and 3 issues proposed for closure (won't fix or follow-up spec).
    *   **Quick and ICE:** The issue of RTP over Quick's interaction with Quick over ICE was closed, asserting that this document defines RTP over Quick with its own ALPN, and other scenarios would be for separate specifications.
    *   **Multiplexing:** The current generic multiplexing framework using a flow identifier might be problematic. An alternative proposed is to define the RTP Quick ALPN for RTP/RTCP only; other multiplexing scenarios would require new ALPNs and separate documents. A pull request will be submitted for this change.
    *   **Congestion Control:** Guidance is provided without being overly specific about algorithms.
    *   **SDP for RTP over Quick:** It was decided that Spencer's existing SDP draft should move towards a working group adoption call, with M-MUSIC review.
    *   **Publication Track:** Discussion favored publishing RTP over Quick as an Experimental or Informational RFC rather than a Proposed Standard, given outstanding questions and ISG requirements for "experiment" definition.
*   **Green Metadata Message Timing Configuration:**
    *   Updates address handling requests for temporal/spatial resolution greater than SDP-negotiated values, focusing on describing receiver behavior.
    *   Security considerations now include scenarios for increased resolution/framerate due to false messages.
    *   The draft is considered ready for Working Group Last Call.
*   **RTCP Feedback Message Timing Configuration:**
    *   A new SDP parameter, `feedback-meantime` (in milliseconds), is proposed to negotiate timing for specific RTCP feedback messages, rather than just general RTCP intervals.
    *   This parameter would define periodicity for regular feedback (e.g., Transport CC) or a minimum wait time for event-based feedback (e.g., PLI, NACK).
    *   An optional `sync-counter` is proposed to synchronize feedback with RTP timestamp changes (frame boundaries).
    *   The working group is encouraged to review the initial draft and provide comments on the mailing list, particularly regarding the different semantics of `meantime` across examples and interaction with RTCP bandwidth constraints.
    *   An IPR reminder was given regarding early patent declarations.
*   **Haptics:**
    *   The draft was revised based on comments, clarifying terminology, adding ISO haptic definitions, and updating text for scalability.
    *   A related Haptic Coding document (23090-D31) will be shared soon.
    *   The authors are seeking reviewers and implementers.
    *   An IPR declaration for this draft was noted as having been made two months prior.

## Decisions and Action Items

*   **Chairs:**
    *   Follow up with Mo regarding `bp9` to clear MISREF status.
    *   Figure out a working group GitHub repository for the J.2K specification and how to host associated large binary test vectors.
    *   Put out a working group adoption call for the `v3c` Viewport and Region of Interest Dependent Delivery draft.
    *   Figure out where JSEP/RTCWeb-related issues should be discussed, given the RTCWeb working group's status.
    *   Discuss with the Area Directors the appropriate publication track for RTP over Quick (e.g., Experimental or Informational) and the definition of an "experiment" for ISG purposes.
    *   Schedule a Working Group Last Call for the Green Metadata draft.
    *   Discuss timing for the Haptics adoption call.
*   **Bernard:** Shepherd the `v3c` draft through the IETF process.
*   **John (InterDigital):** Post links to the public `v3c` implementation and demo on the mailing list.
*   **Dan R.:** Finalize `Skip` v9 with minimal changes to address Eric's ISG 'discuss' position, aiming to clear it before the March ISG board changes.
*   **Stefan:** Add a comment to the J.2K GitHub issue regarding the use of open-GOPs in web environments.
*   **Mat (RTP over Quick Author):** Propose a pull request to restrict the RTP Quick ALPN to RTP/RTCP only, indicating that other multiplexing scenarios require new ALPNs and separate documents.
*   **RTP over Quick Authors:** Create a pull request to ensure consistent use of the term "datagram" (Quick vs. UDP datagrams) throughout the document.
*   **Spencer & Victor:** Collaborate on the SDP for RTP over Quick draft and prepare for a working group adoption call.
*   **Working Group (General):**
    *   Review `v3c` draft, especially from RTP experts.
    *   Review `J.2K` draft, especially from RTP experts.
    *   Review Appendix B of the RTP over Quick draft (RTCP analyses).
    *   Review the RTCP Feedback Message Timing Configuration draft and provide comments on the mailing list.
    *   Review the Haptics draft and provide comments on the mailing list.
*   **Chairs (Internal):** Open an issue on "next steps" for RTP over Quick, listing topics not covered by the current draft but are of interest for future work.
*   **Spencer (Note Taker):** Consolidate notes from `d01` into `d00` for final minutes.

## Next Steps

*   All working group members are encouraged to review the drafts mentioned above and provide feedback on the mailing list, particularly for `v3c`, `J.2K`, RTP over Quick (Appendix B), and RTCP Feedback Message Timing Configuration.
*   The chairs will work on scheduling Working Group Last Calls for ready drafts (`Green Metadata`) and adoption calls for new proposals (`v3c` Viewport/ROI, `Haptics`, `SDP for RTP over Quick`).
*   Discussions regarding the publication track for RTP over Quick (Experimental vs. Informational) will be held with the Area Directors.
*   Efforts will continue to resolve outstanding ISG issues for `Skip` to facilitate its progression.
*   The AVTCORE working group will continue to monitor and coordinate with relevant groups for `HEVC RTP for WebRTC` and `JSEP` issues.