Rilo
Rilo is a story-first vertical video generation pipeline.
It turns a story into script, narration, keyframes, segments, and a final composed video with optional subtitle alignment and burn-in.
Quickstart
npm install -g @telepat/rilo
rilo settings
rilo --project demo --story-file ./story.txt
If you prefer environment variables, export them before running:
export RILO_REPLICATE_API_TOKEN=...
export RILO_API_BEARER_TOKEN=...
If you're running Rilo from source for development, see /contributing/development for npm run dev and npm run dev:all workflows.
If you prefer API-driven runs from a checked-out repository, start the HTTP server with npm run api and check:
- Swagger UI at
/docs - OpenAPI JSON at
/openapi.json
Sections
- Getting Started: install and first run
- Guides: pipeline workflows and operation
- Reference: CLI, env vars, models, artifacts, API auth/webhooks
- Technical: architecture and run-state internals
- Contributing: development and docs/release process
Recommended Reading Path
Story In, Video Out
Start from a single story and let Rilo generate script beats, keyframes, video segments, narration, and a final composed cut.
Model-Configurable Pipeline
Swap models per generation stage and tune each stage with validated model options through project configuration.
Built for Operations
Use checkpointed runs, targeted regeneration, API auth, webhooks, and backend sync to run production-friendly workflows.