What is the Unified Canvas, and when should I use it?

Invoke’s Unified Canvas is designed to streamline and simplify the process of composing an image using generative AI, offering control and composition capabilities within a single image-based editor. It offers artists all of the available generation modes (Text To Image, Image To Image, Inpainting, and Outpainting) in a single unified workflow.

The Canvas allows you to tweak and edit image generations using a brush and eraser, extend images beyond their initial size, and to reimagine new content in a freeform way both inside and outside of existing images.

This document explains the basics of using the Unified Canvas, introducing you to its features and tools one by one. It also describes some of the more advanced tools available to power users of the Canvas.

Overview

Bounding Box

The Bounding Box determines where new images are processed and applied, and is represented by the rectangular selection area on your Canvas. Control its size and location by using the Move tool (V). Based on what is contained in the Bounding Box, different generation methods will be used: Inpainting, Outpainting, ImageToImage, or TextToImage.

Inpainting & Outpainting

"Inpainting" refines parts of an image by using the Mask tool, while "Outpainting" expands the image beyond original borders. If the Bounding Box contains any masked content inside of it, Invoking will Inpaint that section. If there is any missing image content (including transparent/erased areas), Invoking will fill those areas in with color information from the original image (using your current "Infill" setting) and then regenerate those areas.

Text to Image

If there is no content inside of the bounding box at all, a new Text to Image generation will be created when Invoking.

Image to Image

If image content exists inside of the bounding box with no missing areas, and there is no masked areas, Invoking will result in that section of the image being passed through the Image to Image process.

Layers 

Unified Canvas is a multi-layered tool: the Base Layer contains image content, and the Mask Layer is used to define and control image generation areas. Toggle between layers with the dropdown menu at the top left corner or by pressing (Q).

Base Layer

This layer contains your current image content. Directly manipulate this layer with the Brush (B) and Eraser (E) tools. Uploading new images will overwrite the existing content.

Mask Layer

This layer includes masked sections for Inpainting generations. Create or edit masks using the Brush and Eraser tools. Masks only affect the current bounding box.

Staging Area

Generated images appear here. You can't interact with the Canvas while using the Staging Area. Review, save, commit, or discard generated images to resume editing the Canvas - Unsaved images will be lost when the Canvas resets.

Advanced Features

Compositing Settings

Compositing settings allow you to control how new generations are composed into existing images, specifically configuring the Denoising edge size, blur, strength, and steps to control for optimal results.

Scaling

"Scale Before Processing" will upscale the selected area to a larger resolution before running the generation process. Note that this may result in increased generation times, even if the target resolution for the area being generated ends up being smaller - You'll see increased details as a result.

Hot Keys

Unified Canvas includes keyboard shortcuts, accessible by hovering over each button to view the Tooltip associated with that button.