flowGen routes each node to a wired provider: images (Freepik/Vertex), video (Veo, Luma, Runway, Kling), avatar (Hedra), voice (ElevenLabs/Google), music (Suno), 3D (Meshy).
Where: node → Model dropdown — Pick a model on a generation node
Key ideas
- Per-type providers: Each node type exposes the providers/models it supports; capabilities (extend, audio, seed, aspect ratios) vary by model.
- Persist results: External CDN URLs expire; flowGen mirrors them to Firebase Storage.
Steps
- Select a generation node.
- Open the Model dropdown and pick a provider/model.
- Run — the matching generator is called and the result persisted.
Capability-driven controls
The selected model controls which handles and settings are legal—reference image/frame, source video, audio, duration, resolution, aspect ratio, or text modes. Changing models can therefore add or remove inputs rather than merely changing price.
Choose by job, not vendor
- Check the input types the workflow already has.
- Match duration/resolution/audio requirements.
- Compare estimated Flowies before running.
- Use a small test before committing to a high-cost or batch run.
Troubleshooting
A connected reference is ignored or its handle disappeared
The current model does not support that input role. Choose a compatible model or adapt the reference to a supported type.