Image Nodes

The Image node generates images from text prompts, transforms existing images using reference inputs, or does both at once. It supports multiple AI models — each with its own strengths — that you can switch between directly inside the node.

What you can do and how to use it

Text to Image — type a prompt and generate an image from scratch.
Image to Image — upload or connect a reference image; the model uses it as a visual starting point.
Multi-reference — some models accept several reference images for style blending or character consistency.

How to use it:

  1. Add an Image node to your canvas from the node picker

  2. Type your prompt in the text field

  3. Optionally connect a File node or another Image node output as a reference input

  4. Select a model from the model picker inside the node

  5. Adjust aspect ratio and number of outputs

  6. Click Generate

Results appear directly in the node. Use the Output Gallery node to compare outputs side by side.

Available models and settings

Available models (check the model picker for the current full list):

  1. Flux 2 Flex / Flux 2 Max (Black Forest Labs) — Fast and flexible to higher quality outputs

  2. Seedream 4.0 / 4.5 / 5.0 (ByteDance) — Strong prompt adherence, clean composition, progressively higher quality

  3. GPT Image 1.5 / 2 (OpenAI) — Versatile, strong instruction-following

  4. Nano Banana / Pro / 2 (Google) — Fast lightweight to improved style coherence

Key settings:

Aspect ratio — controls output dimensions: 1:1, 16:9, 9:16, 4:3, 3:4. Available ratios depend on the selected model.

Number of outputs — generate multiple variations in a single run. Maximum varies by model.

Reference images — the number of reference images you can attach depends on the model. More references generally improve style and character consistency.