Carry bounded H3 temporal context through chained video delivery
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 5/5
- Estimated time
- Over a week
- Newbie friendliness
- 25/100
- Issue type
- Feature
- Clarity
- Clearly specified
- Activity status
- Active
- Tech stack
- javascript, node.js
Research direction
Read modeContract.js, prepareParams.js, renderArgs.js, chainedVideo.js, and continuousVideo.js, then trace queue validation and model capability projection. Start by confirming which runtime pins support verified temporal continuation; the issue also depends on #13668, #13669, and #13670. Done means the admission and render paths agree, three successive chunks preserve bounded frame and audio history, and replay records the requested and effective strategy plus refusals.
Written by the indexing model from the issue text.
Description
Current evidence: chainedVideo.js takes source-video tails only for LTX window hops; continuousVideo.js backendJobParams converts local handovers into image mode. Queue replay and helper arguments do not express verified H3 temporal history.
Chosen fix: use the runtime contracts and backend adapters to add explicit temporal continuation admission with a last-frame opt-out. Wire requested strategy and bounded context through modeContract.js, prepareParams.js, renderArgs.js, chainedVideo.js, continuousVideo.js, queue validation/provenance and model capability projection. Refuse an explicitly requested temporal window on an unverified runtime, never silently report success after falling back. Preserve existing requests that implicitly use legacy last-frame behavior and keep source media intact.
Acceptance criteria:
- Admission before queue persistence and render/replay boundary agrees on supported pins and explicit refusals.
- Three successive chunks deliver each new frame exactly once; bounded history does not grow with the episode, repeat prefix frames, lose seam frames or repeatedly re-encode the accumulated source.
- Tail video and corresponding audio intervals reach the correct adapter; timing is runtime-specific.
- Cancel/restart/history-preparation failure removes only owned scratch assets, preserving completed clips and original source.
- Persist requested/effective strategy, actual history length, backend pin/capability and refusal reason for later disclosure; preserve LTX window and source-splice coverage.
Scope: existing videoGen admission, queue, render and chain services, model capability projection and public-boundary tests.
Dispatch: heavy capability for cross-service async lifecycle and frame/audio delivery correctness; xhigh effort for persisted replay, failure and three-chunk invariants.
Blocked by #13668, #13669, #13670
Part of #13614
After shipping, check this slice in #13614. If this is its last open child, post a summary of all delivered children and close #13614; otherwise leave the epic open.
- Dominant language
- JavaScript
- Stars
- 38
- Forks
- 32
- Avg merge
- 44m
- Merged PRs (30d)
- 990
Getting set up
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from atomantic/PortOS
-
area:media blocked cuda effort:high enhancement help wanted model:medium plan windows
Difficulty 4/5 3-5 days Newbie friendliness 35/100
atomantic/PortOS#13683 · 1 comment ·
Maintainers usually reply within 1 day
-
apple-silicon area:media blocked effort:high enhancement help wanted macos model:heavy plan
Difficulty 5/5 Over a week Newbie friendliness 20/100
atomantic/PortOS#13682 · 1 comment ·
Maintainers usually reply within 1 day
-
area:media blocked cuda effort:medium help wanted model:medium plan windows
Difficulty 3/5 1-2 days Newbie friendliness 35/100
atomantic/PortOS#13681 · 1 comment ·
Maintainers usually reply within 1 day
-
apple-silicon area:media blocked effort:medium help wanted macos model:medium plan
Difficulty 4/5 3-5 days Newbie friendliness 28/100
atomantic/PortOS#13680 · 1 comment ·
Maintainers usually reply within 1 day
-
area:media blocked effort:high model:medium plan planner:gpt-6-1-sol
Difficulty 4/5 3-5 days Newbie friendliness 35/100
Maintainers usually reply within 1 day
All issues in atomantic/PortOS
Similar issues
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
glossyio/traffic-monitor#255 ·
Maintainers usually reply within 1 day
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
juice-shop/juice-shop#3706 ·
Maintainers usually reply within 1 day
-
Main workflow runs twice for newly created refsPossibly taken @rustyconover claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
duckdb/duckdb-wasm#2293 ·
Maintainers usually reply within 1 day
-
Difficulty 1/5 Under an hour Newbie friendliness 85/100
521xueweihan/HelloGitHub#4115 ·
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
outline/outline#14348 · 1 comment ·
Maintainers usually reply within 1 day