Image nodes

Image nodes generate, hold and edit still pictures, and they're usually the first step before video. When you select an image node, a toolbar appears above it with the most-used image editing tools.

Create an image node

• Choose “Image” in the “Add node” menu. By default it generates an image from text;
• Connect an image to it and the node switches to image editing (image to image), building on that picture, and the model list changes to match;
• In the prompt bar's settings you can switch “Generation method” from “Image generation” to “Layer decomposition” to split a picture into layers you can edit separately (paid);
• You can also upload, drag in or paste an image as an asset node to use in later generations.
• When a node has several result images, expanding it shows them in a 2×2 grid; with more than four, page left and right.

Image editing tools

With an image node selected, these are the main tools on its toolbar (with whether each is billed):

• Crop: drag the handles to trim the picture, with preset or custom ratios (free);
• Multi-angle: drag the 3D cube to pick a new camera angle, tilt and zoom, and generate the same scene from a different viewpoint (paid);
• Redraw: brush or box-select the area to change, describe what it should become, and the AI regenerates that area (paid);
• Relight: drag the light sphere to set the direction, strength and color temperature of the key and rim lights, and regenerate the lighting (paid);
• Panorama: turns the current image into a 2:1 panorama (paid);
• Rotate & mirror: rotate by a preset or any angle, and mirror horizontally or vertically (free);
• More → Layer decomposition: splits the image into layers you can edit separately. It creates a new node downstream, where you confirm the cost and submit (paid). Once a node has layers, “More” also has “Edit layers” (combined on your device, free);
• More → Outpaint: drag the frame outward to extend the picture, choose a ratio and resolution, then generate (paid);
• More → Erase: circle what you want removed and generate, with no description needed (paid);
• More → Markup: draw colored marks on the image with a brush or box and save the result as a new image (free);
• More → Photo enhance: AI HD redraw, with a choice of model and target resolution (paid);
• More → Resize pixels: enter a target width and height (you can lock the ratio) to resize directly (free);
• More → Remove background: keeps the subject and removes the background (paid);
• More → Quick split: splits the image into a 2×2, 3×3 or 4×4 grid. Hover over this row or click the arrow on its right to open a grid picker and choose your own rows and columns. The total number of tiles has a limit, and a split over it can't be submitted (free);
• Create element: an image with a result can be saved as a reusable element (character, location or prop) that you then @-mention in prompts (free).

When a node has two or more connections, “Disconnect all” also appears at the start of the toolbar.

Panorama preview

Once there's a result image, you can open a WebGL 720° spherical preview: drag to rotate and scroll to zoom. You can also capture 4 or 12 views at once and send them back to the canvas as reference images (free).