Generate Your First Video
Banana Flow’s video models can take either a text prompt or a starting image (image-to-video). Most produce 4–12 second clips at multiple aspect ratios.
1. Add a video node
From an empty board, add a Video node.
2. (Optional) Add a starting image
For image-to-video, drop an Image node on the canvas, upload (or generate) a frame, and connect it into the video node’s images target handle.
You can also skip this step entirely and let the model generate purely from text.
3. Add a prompt
Add a Prompt node, write what should happen in the clip — camera movement, action, mood — and connect it to the video node’s prompt handle.
A slow dolly forward through morning mist over a forest floor; soft golden light filters through the trees.
4. Pick a model
In the video node, choose a model:
- Veo 3 — Google’s flagship; strong camera control and physics
- Seedance 2.0 — multimodal directing, native lip-sync audio, up to 12s
- Kling 2.1 — excellent character motion
- Runway Gen-3 — established workhorse
See Video Models for the full lineup, durations, and per-generation cost.
5. Generate
Click Generate. Video generations take longer than images (typically 30s–3min depending on model and length). The credit cost is shown on the node before you commit.
The finished clip appears in the node and is stored in your Media Library.