Repository navigation
Plan and preview one timed generation for a recurring music-video section - #14447
Merged
Merged
Conversation
…14241) Add the pure planner for sequence coverage: group a section's cutaway footage shots that share a set and protagonist into a single reference-to-video request, compose the timed-range prompt with each model's own image naming, and map every scene back to its existing in/out inside the returned clip. Performance shots, mismatched sets, gaps and runs over the model max stay per-shot; Kling stays per-shot (no verified multi_prompt body). buildFalVideoRequest accepts a fixedOverrides entry so a sequence disables H3 prompt expansion.
Scenes accept coverage: 'sequence'; GET /api/music-video/:id/sequence-coverage previews the one-request runs, the exact prompt and request fields, and each scene's cut inside the clip. Read-only; nothing is submitted.
atomantic
added
enhancement
This was referenced Oct 31, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Refs #14241. This is the planning slice; the issue stays open.
What ships
server/lib/musicVideoSequenceCoverage.js, pure): groups acoverage: 'sequence'section's cutaway footage shots (same set + protagonist, contiguous, no performance shot, inside the model max) into one-request runs, and records why every other shot stays per-shot. It composes the prompt as contiguous time ranges, naming references the way each model expects (Image non H3,@Imagenon Seedance), and maps each scene back onto its existing in/out points in the returned clip. No new cut times are invented. Kling stays per-shot because no verifiedmulti_promptbody exists.coverage: 'per-shot' | 'sequence'is in the create and update schemas,buildScene, andsplitScene, with a parity test.GET /api/music-video/:id/sequence-coveragereturns the grouped runs, the exact prompt and fal body (H3 sendsprompt_expansion_mode: 'disabled'through main'spromptExpansionoption), the labeled prompt-expansion addition, and the clip spans. It submits nothing and spends nothing.Still open on #14241
falSceneTakecalls.Tests
Planner tests cover the split, the performance exclusion, the over-max refusal, and the H3 expansion value. A service preview test covers the same three-shot case. No provider calls.
npm run pregatepasses.