Image Node
Upload images or generate them from a chat in Image mode, then iterate and chat about them on the canvas.
Last updated: July 19, 2026
The Image Node holds a picture on your canvas — one you upload, or one generated from a Chat Node in Image mode. Either way, the image's content is parsed so connected chats can discuss it.
Supported files
- JPG, PNG, GIF, WebP
Uploading an image
- Double-click the canvas and choose the image icon
- Click Add From Computer, or drag and drop an image file onto the node
- The AI parses the image content automatically, so any connected chat can discuss what's in it
After parsing, open the Parsed Content dialog on the node to see what the AI extracted.
Generating images
Image generation happens from a chat node switched to Image mode (⌘3):
- Switch a chat node to Image mode — the model picker swaps to the image model catalog ("Search image models..."), with a fast, affordable default (currently Gemini 3.1 Flash Image) already selected
- Describe the image you want and send
- Each result branches into its own connected image node, so the conversation stays readable while the visuals live on the canvas
Aspect ratio, resolution, quality, and format are conversational — ask for "a wide 16:9 hero image" or "a transparent PNG" and the settings are applied for you, within what the chosen model supports. One prompt produces one image, and the chat never locks while a generation runs.
Image generation requires a paid plan. Prefer your own billing? Turn on Use my OpenRouter key for image generation in Settings → Providers → OpenRouter — generations then run on your OpenRouter account instead of your credits, and those images are captioned "via your OpenRouter key".
Editing and variations
- Edit button — a finished generated image gains an Edit action. Click it and a new Image-mode chat spawns already wired to that image, ready for "make the sky dusk" follow-ups.
- Reference images — connect any image node (uploaded or generated) into an Image-mode chat and it becomes a candidate reference. Say what you want done with it and the next generation uses it for edits and variations; an image connected purely as context is left alone.
Image nodes only accept other image nodes as manual inputs — text, web, and document nodes can't be wired in as generation references.
Chatting about an image
Connect an image node into any chat (or @-mention it) and the Parsed Content — the AI's description of the image — flows in as context, so the conversation can reference what's actually in the picture.
Use cases
- Research — drop in charts or diagrams from papers; the parsed content lets a connected chat explain or compare them.
- Writing — generate a mood or scene image to anchor a piece of descriptive writing.
- Studying — upload textbook figures and ask a connected chat to walk you through what they show.
- Content creation — iterate on thumbnails with the Edit button, keeping every variation on the canvas for side-by-side comparison.
- Planning — sketch or screenshot a layout, upload it, and have the AI critique or list next steps.