Framesail docs
How to use Framesail, task by task: set up a channel style, keep characters consistent, and take a project from script to finished MP4. Every station of the pipeline is editable and re-runnable, and these guides show where each control lives.
Getting started
What Framesail is, how the pipeline fits together, and your first video.
- How Framesail worksThe pipeline from script to finished video — every station, what it produces, and how the steps unlock in order.
- Organize channels, projects, and settingsHow channels, styles, and projects relate, plus every project-level setting and what it controls.
- Create your channel styleSet up a style by analyzing references or picking presets, then finish its character and environment template images.
- Make your first video, end to endA walkthrough of the five in-app steps, what unlocks each one, and where you can iterate along the way.
Styles & consistency
Define a look once and hold it across every shot of every video.
- Analyze reference videos into a styleWhat style analysis reads, the editable fields it produces, and how re-analysis works without losing your edits.
- Choose style presetsThe curated preset catalog — art, narrative, and director styles — mixed one per axis.
- Edit style fields and add custom variablesThe built-in style fields, the real-media switch, and @variables you can wire into specific generation prompts.
- Keep characters consistent across scenesHow locked reference images keep characters, environments, and objects on-model in every shot.
Pipeline steps
Every station in depth — controls, models, and regeneration.
- Script: generate, paste, or reviseThree ways to get a script into a project — generate, paste, or revise — plus length control, dialogue formats, and versions.
- Script scan: your cast, world, and voice blocksWhat Analyze Script extracts, why re-running it is destructive, and the non-destructive voice-block re-extraction alternative.
- Reference images: generate, upload, regenerateGenerate or upload one locked reference image per character, environment, and object — the step that unlocks voiceover.
- Voiceover: voices, providers, speed, and pausesTTS providers and voices, narrator and character voice binding, speed ranges, break tags, and safe batch regeneration.
- Storyboard: the AI director's planHow the storyboard is planned against real narration timing, plus focus tracks, continuation, segment operations, and real media.
- Segment renders: images, clips, and modelsImage and video model choice per run, per-model settings, prompt overrides, and per-shot regeneration rules.
Editing & export
The editor, captions, audio, and the finished video.
- The editor: timeline, scenes, and layersScenes on an absolute timeline, layer editing, the per-scene AI director panel, and what the editor deliberately is not.
- CaptionsTurn captions on in the editor and style them — presets, fonts, layout, colors, and animation, timed from your voiceover.
- Music & sound effectsThe shared audio library, user-selected music tracks with volume and trim, per-segment SFX, and the loudness policy.
- Export & downloadOne-click cloud export to a flat, watermark-free 1080p MP4 — what it requires, what it costs, and what you download.
Account, billing & API
Plans, credits, API keys, MCP, and job status.
- Plans & creditsWhat Creator, Pro, Studio, and BYOK each unlock, plus how credit grants, rollover, and expiry behave.
- API keys & the REST APICreating and revoking fsk_ keys, which plans include API access, and what the REST API surface covers.
- MCP server & connecting agentsConnect Claude or any MCP client to Framesail — API-key and OAuth paths, and the server-side guardrails agents get.
- Bring your own keys (BYOK)Which providers you can connect, how per-job coverage bills 0 credits, and how keys are stored.
- Jobs, errors, and statusHow background jobs report status, why retries never double-bill, and what step-order errors mean.
Building with the API or MCP instead? The technical reference lives at API & MCP docs.