The Image Generator node runs an AI image model and produces one or more images as output. You can connect it to a Request Input to make your prompt dynamic, chain it with an Assistant to craft prompts automatically, and wire its output to a Response Output to expose the result through your workflow’s API.
Available models
The model picker on the node lists every enabled image model. Three models are available by default:
- Nano Banana 2 — a fast, general-purpose model suitable for most image generation tasks
- Seedream V5 Lite — high-quality model with more detailed output
- GPT Image 2 — OpenAI’s image model accessible via the registry (Business+ tier)
Your workspace administrator can add more models through the Node Builder. The picker always reflects what is currently enabled in your workspace.
The inputs available on a given node depend on the selected model. Common inputs include:
| Socket | Type | Description |
|---|
prompt-in | Text (orange) | The generation prompt. Connect a Request Input or Text Box here. |
image-in | Image (pink) | General reference image |
style-in | Image (pink) | Style reference image |
character-in | Image (pink) | Character reference image |
start-image-in | Image (pink) | Starting image for image-to-image generation |
end-image-in | Image (pink) | Target image for morphing workflows |
params-in | Any (purple) | Parameter bundle from a Request Input node |
Output sockets
The image-out socket on the right side of the node emits the generated image. Connect it to a Response Output, another generator, or an Assistant node for further processing.
Controls on the node
Below the prompt textarea you will find two rows of controls:
First row
- Count stepper — how many images to generate per call (1–15). Disabled when a List node is connected upstream, because the list length controls the batch size in that case.
- Model picker — select the active model.
Second row
- Aspect ratio dropdown — choose from the options the selected model supports (for example, 1:1, 16:9, 9:16).
- Resolution dropdown — appears when the model exposes resolution presets.
- Settings icon — opens the side panel for advanced configuration.
- Run button — run the node.
Credit cost
The estimated credit cost for the current settings is shown in the node header (for example, ~35 cr). The cost updates when you change the count or, for models with per-second pricing, when you change duration.
Side panel inspector
Click the Settings icon to open the side panel. From there you can:
- Upload reference images (style, character, start/end) with large tile pickers
- Edit the prompt in a full-width text area
- Adjust advanced parameters the model exposes (negative prompt, guidance scale, seed, and others depending on the model)
- Choose output format
Changes in the side panel sync instantly to the node and vice versa.
A simple image API workflow is: Request Input (prompt field) → Image Generator → Response Output. The caller sends {"prompt": "..."} and gets back an image URL.
Generation history
Each time you run the node, the new result is appended to the node’s generation history. Use the arrow controls below the preview to navigate between past generations. Click All to open a full-screen gallery where you can browse and delete individual results. 