System

Settings

Providers, models, clip defaults and pipeline tuning.

Video Generation

AI video generation. Story uses your LLM Tasks; the media steps use fal / Replicate (images can also use OpenRouter, or free Cloudflare Workers AI / Pollinations / Free AI). Set FAL_KEY / REPLICATE_API_TOKEN in .env; OpenRouter reuses OPENROUTER_API_KEY; Cloudflare needs CLOUDFLARE_ACCOUNT_ID + CLOUDFLARE_API_TOKEN; Pollinations takes no env key but bills credits (Bring Your Own Pollen). These providers are text-to-image only (no character-sheet consistency). Leave a model blank to use the provider default.

Enable video generation

NichesComma-separated.Character sheet — provider

Character sheet — modelReference image for character/style consistency. OpenRouter e.g. google/gemini-2.5-flash-image; Cloudflare e.g. @cf/black-forest-labs/flux-1-schnell; Pollinations e.g. flux.Scene image — provider

Scene image — modelPer-scene image, conditioned on the character sheet (paid providers only). Free: Cloudflare @cf/black-forest-labs/flux-1-schnell, Pollinations flux.Voiceover (TTS) — provider

Voiceover (TTS) — modelNarration voice. Left on fal/Replicate, non-English narration is still routed to free Edge voices unless a video pins its own generator.Voiceover (TTS) — voiceProvider-specific voice id. Blank = the model's default voice.Music — provider

Music — modelBackground music bed.AI video provider

AI video modelDirect text/image-to-video for Video Mode. fal e.g. fal-ai/ltx-video; Replicate e.g. lightricks/ltx-video; Pollinations (BYOP) e.g. wan-fast / seedance / veo.