Settings

Providers, models, clip defaults and pipeline tuning.

General Providers Tasks Prompts & Agents Clip Defaults Transcription Processing Rendering Moment Scoring Storage Video Generation

Moment Scoring

How the LLM judges candidate moments. Fewer candidates / smaller batches = cheaper.

Max candidatesMoments sent to the LLM after the local pre-filter.Batch sizeMoments judged per request (fewer = cheaper).Parallel requestsLower to avoid provider rate limits.Context padding (seconds)Transcript context around each candidate.