The meeting focused on two main technical presentations: David Hassoun’s presentation on extending DASH 6th edition to support non-linear ad formats like overlays and multi-view presentations, and Bertrand’s discussion on reducing live manifest parsing time using X-Link. David demonstrated how to implement custom layouts using CSS viewport approaches and IAB specifications, though concerns were raised about synchronization and implementation complexity across different devices. Bertrand shared his findings on implementing on-request X-Link mode, which significantly reduced parsing time from over one second to 60 milliseconds on low-end devices by only resolving necessary periods when needed. The group discussed challenges with X-Link implementation, including caching behavior and timing specifications, with Alex and Thasso expressing concerns about the approach based on previous implementation issues. The conversation ended with plans to continue both discussions in the next meeting, including reviewing Bertrand’s implementation details and exploring alternative solutions using MPD insertion events.
1. Alex – Prepare an example using MPD insertion event (similar to Bertrand’s XLink approach) for reducing manifest size/parsing time, to be presented and discussed at the next meeting.
2. Bertrand – Send presentation slides and GitHub link with Dash.js implementation of XLink on-request to Thomas for distribution to the group.
3. Daniel – Review David’s proposal and implementation approach for overlay/non-linear ad support in DASH, focusing on test implementation aspects.
4. Daniele – Prepare and submit a small document detailing the test results of legacy player behavior with LCBC non-scalable proposal sub-representations, and bring it for review at the next meeting.
5. Daniele – Test the LCBC non-scalable proposal with Video.js player and include results in the document.
6. David – Share the presentation slides and relevant documents with the group, and prepare to revisit the topic in two weeks for further discussion and feedback from implementers (especially Daniel).
7. Thasso – Continue work on JSON-MPD documentation, focusing on editorial and website preparation, and be prepared for further review when ready.
8. All interested parties (especially Thomas, Daniel, Alex, Thasso, Iraj): Review Bertrand’s XLink on-request proposal and Alex’s upcoming insertion event example before the next meeting to facilitate technical discussion.
- Antitrust Statement & Project Update and Release Planning
- The meeting began with roll call and agenda setting, where Thomas announced that David would present.
- Thomas mentioned a discussion with DVB colleagues about evaluating the 6th edition of DASH, though this topic was postponed.
- Daniel provided an update that a planned release was scheduled for the following Thursday, and Daniele requested time for an update on the LCBC non-scalable proposal.
- The conversation ended with approval of the agenda and previous minutes, with plans to proceed to David’s presentation.
Manifest Update for Ad Presentations
- David presented a proposal to update the manifest to support both linear and non-linear ad presentations, including overlay formats, building on previous discussions at the SVTA face-to-face meeting.
- Nicolas outlined requirements and considerations for implementing this approach, including handling devices with single or multiple decoders, and demonstrated examples of custom layouts using CSS viewport approaches.
- The team discussed the need to balance supporting complex layouts while maintaining compatibility with existing devices, with Thomas suggesting they should gather feedback from implementers like Daniel before proceeding further.
Ad Overlay Rendering Specifications
- The team discussed the accuracy and implementation of rendering instructions for ad overlays, with David explaining that the current specifications are fairly generic, providing guidelines for corner overlays that are approximately 25% of the screen size.
- Thomas proposed creating a descriptor with a schema and value that points to the IAB spec to allow clients to implement layouts based on the specification.
- Olivier clarified that there are two IAB specifications: one for creating ads and another for ad signaling, which includes placement and position metadata using two numbers to define different ad formats.
Concurrent Content Display Implementation
- The group discussed implementing concurrent content displays, with Thasso expressing concerns about synchronization when using multiple overlay elements within a single period.
- David clarified that multi-view content is not meant to be synchronized, and Thomas suggested using independent events for overlapping content rather than mixing them in a single period.
- The discussion highlighted a key challenge of supporting both linear and nonlinear content without requiring separate streams or manifests, with Iraj preparing to address different implementation approaches.
DASH Client Rendering Discussion
- David proposed handling all rendering and combination within the DASH client rather than in the application, though he noted no final decisions had been ratified yet.
- The discussion covered technical aspects including volume control for individual assets in multi-view and overlay scenarios, with David explaining that volume control per asset would allow for audio prioritization and ducking capabilities.
- The meeting was cut short due to multiple participants needing to drop, with Thomas indicating that LCEVC, DRM DVB news, dash.js v5.2.0 introduction, Bertrand’s presentation, and AOB items were still to be covered.
Video Overlay Implementation Challenges
- Alex discussed complications with overlay implementation, specifically mentioning transparent video use cases using either separate videos with transparency or enhancement layers in scalable video codecs like AV1 and LCEVC.
- Daniele from V-Nova explained that implementation methods vary by device, including using Canvas for web players or pipeline annotations for GL or Vulkan pipelines, with enhancement layers potentially having different resolutions than the base layer.
- The discussion was cut short when Thomas needed to leave, with Yuriy taking over the meeting.
Presentation and Testing Progress Update
- The team discussed next steps for a presentation, with Thomas suggesting Daniel review the test implementation and proposing to reconvene in two weeks.
- Thasso provided an update on JSON-MPD documentation, indicating it was complete and ready for website publication with only minor editorial changes needed.
- Daniele reported on testing legacy players with sub-representations, finding no playback regressions across multiple platforms including desktop, mobile, and browser players, though the team agreed to expand testing to include video.js and submit findings as a contribution document.
- Bertrand mentioned a proposal about reducing live manifest size using X-Link, but the discussion was postponed to the next meeting due to missing key participants.
X-Link Implementation Challenges Discussion
- The team discussed the implementation and challenges of using X-Link’s “on-request” mode for live content, particularly focusing on parsing time optimization.
- Bertrand presented his implementation in dash.js which showed significant parsing time reductions from over one second to 60 milliseconds for large manifests with long DVR periods on low-end devices.
- The group identified several issues with the approach, including caching behavior problems, unclear timing specifications in the W3C spec, and concerns about MPD updates returning to X-Link behavior.
- The team agreed to continue the discussion next week, with plans to review Bertrand’s presentation and explore alternative solutions including JSON manifests and MPD insertion events as potential alternatives to X-Link.
Next Meeting: Tuesday, May 26, 2026 9 am PDT
Chat Record:
- David Hassoun shares: Example Layout Tester
https://www.svta.org/wp-content/nlag/v4
Proposed DASH Approach
https://docs.google.com/document/d/1CWm1BP65h45iJ5RwCAQkUAZ6lz7f-w3qJRGMdH8DDdQ/edit?tab=t.x2hudn1ltnpd
SVTA Signaling Specification (needs updating)
https://docs.google.com/document/d/16CZO5Q2WXwXOK7pVQm03crS0W20l8LxP/edit?pli=1#heading=h.2jxsxqh