Inpainting

Image+text-2-image

Inpainting

Using a start image you can create a new AI artwork based on the composition of your original image

Quick menu: click on any section to quickly go to it

Key features

Step by step tutorial

Parameters description

Key features

Change parts of image
Use any image and mask to change parts and add new objects
Use generations or photos
You can use it with any generated image or already existing photo or picture
Works as Outpainting
With this node you can also expand the image by combining it with Image Unzoom node
Basic and advanced settings
Tweak the detection of mask, its blur and other parameters

How to use the node

Click on each step to see the corresponded image as an example and tutorial to follow
Important note: Inpainting doesn't work with Trained models

1. Hover over the image and press Inpaint
You can do it with any generated image or uploaded image/photo. Simply hover over it and press Inpaint
2. Paint over the area you want to change, type prompt and press Generate
The window to paint will appear in the upper right corner, simply paint with white over the area you want to change.
In prompt you should write what you want to see instead of white area. For example, if you want to change hair, you should type what hair you want to see. Or if you wanted to change clothes, type what kind of clothes you want to see

After your press Generate the nodes will be added automatically to the workspace
3. Check the result
If you don't like the results, simply press Start button again to get different images

1. Hover over the image and press Zoom out
You can do it with over generated image in SD 1.5 or you can manually add node 'Image Unzoom'
2. (Optional) Unzoom the image as you want
Place your cursor over the image in Image Unzoom and zoom out until you find the right resolution.
Click Fixed Zoom in parameters to prevent image from changing
3. Launch the node or type what you want to see
If you simply clicked on Zoom out in SD 1.5, relaunch the node
If you simply clicked on Zoom out in SD 1.5, relaunch the node
If you're using your own image, in prompt you should write what you want to see outside of your image. Type also this negative prompt: tiled, straight line, collage, separated, frame, framed, text, letters, tape, rectangles
If you're using your own image, put parameter Pixels around mask to 10

Parameters

The parameters are all similar to the general Stable Diffusion parameters (for reference, visit this page) . The main parameter that you might want to change is Pixels around mask

Range: from 0 to 200

It defines whether AI should generate content outside of drawn mask. If you want an object to appear bigger and be blended more smoothly, keep it around 10-15.