​

Image Generation

The Image Generation node creates images from text prompts using your choice of 4 AI image generation engines. The default is Nano Banana 2.

​

Inputs

The node also accepts an optional image input for reference-based generation.

​

Multiple reference images

Some engines accept more than one reference image at once. When you enable image input on one of these engines, you can add numbered reference slots — “Image 1”, “Image 2”, and so on. Each slot is a separate connection, and the number stays fixed no matter how you arrange the canvas. To point at a specific reference inside your prompt, insert an [Image N] tag (a chip toolbar above the prompt makes this one click). For example, with a body shot wired into Image 1 and a face wired into Image 2:

Replace the face of the person in [Image 1] with the face from [Image 2], keeping the original pose, lighting, and background.

Because the tag is tied to the slot — not to the order the images happen to be processed in — the reference is reliable every time the flow runs.

​

Outputs

​

Available engines

​

Configuration options

​

Advanced options

​

Credit cost

Varies by engine — see table above. Images are generated at 9:16 aspect ratio by default.

​

Tips

• Prompts are automatically enhanced with cinematic lighting, camera angles, and style keywords
• Use reference images (via the optional image input) for more consistent character generation
• Grok Imagine and GPT Image 2 are the most cost-effective options
• GPT Image 2 is best for in-image text rendering (signs, captions, logos)
• Nano Banana Pro produces the highest photoreal quality

​

Example use cases

• Generating scene backgrounds for video creation
• Creating character images for motion transfer
• Producing cover images or thumbnails for short-form posts
• Creating consistent visual styles across shots