The Edit Image node is where you may edit an image input. This is one of the most commonly-used nodes on Layer, useful in most image editing steps and even for adding UI elements to scenes. Prompt Editing is powerful and flexible when used with a reasoning model like Gemini 2.5 or Gemini 3, and can accomplish many of the goals that make AI generation feel magical.
The Edit Image’s outputs will always be in a 2D format (typically PNG) and contain two types of inputs:
The Images to Edit are the specific base image that you want to have modified. This may be a piece of key art, an object or item, or something like a banner ad background.
The Reference Groups are a collection of reference images that are sent along with the Images to Edit, and may be referenced in the prompt.
You may create up to seven Reference Groups per Edit Image node, which allows up to 7 reference images to be sent along with a single Image to Edit. You are not charged extra units regardless of how many reference groups you send.
When using the Edit Image node, the number of Images to Edit as well as the number of distinct reference images per group will determine how often it runs. For example:
When sending in four Images to Edit, a prompt edit node will run four times.
When sending in four Images to Edit and a Reference Group with a single image, the prompt edit node will run four times.
When sending in four Images to Edit and a single Reference Group with two images, the prompt edit node will run eight times (4 base images x 2 reference images in a single group).
When sending in four Images to Edit and three Reference Groups with one image each, the prompt edit node will run four times 4 base images x 1 reference image in 3 groups).
The images you want to edit (also known as base images) are defined here. The images may either be uploaded directly to the node (i.e. hardcoding the base images) or sent in from other nodes.
You may upload an image directly to the Images to Edit here.
You may select the model you'd like to use here. As of this article's publishing, we recommend using Gemini 3 as the most powerful model that can provide the best results over a wide variety of asks.
You may add your own prompt into the generator or connect prompt inputs to it in the workflows graph.
You may select a specific aspect ratio, but for most cases we recommend using a dynamic aspect ratio, especially if multiple aspect ratios of images are being sent into the node.
Reference Groups define the number of reference images you want to send in. You should add one group per additional reference image you want to send in, up to 7 groups.
Please note that if you send in multiple reference images within a single group, you will run the node that many times. For example, sending in 5 reference images within Group 1 would run the Prompt Edit node 5 * (# of Base Images).
Instead, if you would like to send 5 reference images per each base image, create 5 reference image groups and send in one image per group.