Use the shared VideoGenerationNode card for two different Add actions: create a new clip with Gen Video, or continue a connected source with Extend Video.
Where: canvas → left rail → Add → Video → Gen Video — Add a Video Generation node and connect a Prompt
Key ideas
- Two visible tasks: Gen Video starts a clip from prompt and supported frames/media. Extend Video requires a continuation-oriented source and brief; it should not be documented as merely another label for fresh generation.
- Model-driven controls: Model capability decides which prompt, frames, duration, aspect, resolution, audio/SFX, person, seed, negative, FPS, and orientation controls appear on the card/sidebar.
- Opened surfaces: Results open in video lightbox and can enter annotation, frame extraction, seam repair, composition, or rendering.
- Gen Video card: VideoGenerationNode uses the authored shared-card path. Generate a versioned video clip from prompt and model-supported image/frame/video/audio inputs.
- Properties sidebar: imageVideoNodes.tsx
- Opened surfaces: Video lightbox; annotation entry
Steps
- Add a Video Generation node and connect a Text Prompt node to its Prompt (text_in) handle.
- Open the node (expand) and pick a model in the Model dropdown — Veo 3.1 is the Google default and the only model exposing 4K and image-reference inputs.
- Set Duration (e.g. 4 / 6 / 8s for Veo), Aspect Ratio (16:9 or 9:16), and Resolution; toggle Generate Audio on if the model supports it.
- Optionally wire a Start Frame and End Frame image, or open Advanced Options to set a seed for reproducible results.
- Click Generate Video; to continue an existing clip instead, flip the toggle to Extend, connect a video to Extend Source, and click Extend Video.
Gen Video controls and presets
- Add path: Flowgen → Add → Video & Motion → Gen Video.
- Card: VideoGenerationNode.
- Properties: imageVideoNodes.tsx.
- Controls: Model; prompt/frames; duration; aspect/resolution; audio/SFX; person/seed/negative/FPS/orientation.
- Opened surfaces: Video lightbox; annotation entry.
FAQ
Why is the Generate/Extend toggle missing on my video node?
Extend is only offered for models that support it (Veo and Luma Ray-2). If you swap to a model that can't extend, the node silently falls back to Generate and the toggle disappears.
Why does my duration reset after changing the model?
Each model has its own valid duration set, and Veo's Extend mode tops out at 7s. When you change model or mode the node clamps any now-invalid duration/resolution/aspect to a supported value so generation doesn't error at submit.