okuha-logo-small-white

Outpainting In Stable Diffusion – Online And For Free

outpainting-in-stable-diffusion-feature-image

If you click on a link and make a purchase, I may receive a small commission. As an Amazon affiliate partner, I may earn from qualifying purchases.
Read our disclosure.

What Is Outpainting?

Outpainting is a technique that allows you to expand images beyond their original borders using artificial intelligence (AI) and text prompts.

With outpainting, you can create new elements or scenes outside the original picture by generating content that matches the art style and colors of the existing image. The more of the original image the outpainting frame covers, the better the match will be.

For example, suppose your original image size is 512px x 512px. You can increase the image size with outpainting (to something like 1024px x 512px) and continue generating visuals while keeping the original content in place.

Outpainting In Stable Diffusion

stable-diffusion-outpainting-example-result
Stabel Diffusion outpainting results using the Anything V5 diffusion model.

To get started with outpainting, you need the following files:

AUTOMATIC1111 WebUI offers outpainting as well as inpainting features. After you’ve generated an image with a diffusion model of your choosing, click the ‘Send to img2img’ button below the generated image to start the outpainting process.

In the img2img tab, locate Script and select Outpainting mk2 from the list.

stable-diffusion-outpainting-script-location
When in the img2img tab, locate Scripts and select Outpainting mk2

The best part about using Stable Diffusion outpainting in AUTOMATIC1111’s WebUI is that you can select any model you have downloaded. This will keep the outpainting process seamless and generates the same aesthetics and visuals for the expanded area.

When sending the newly generated image to the img2img, you can continue the outpainting process further.

stable-diffusion-outpainting-example-result-img2img
Sending the left side image to the img2img generator and generating the image again, you can see how the left image is expanded while keeping the image aesthetics intact.

Outpainting Script For Stable Diffusion

Suppose you don’t have the Outpainting mk2 selectable in your Stable Diffusion script tab for some reason. You can get the outpainting script for AUTOMATIC1111 from Github.

When you’ve downloaded the file, copy and paste it to the following folder:

stable-diffusion-webui\scripts

Outpainting In Photoshop With Stable Diffusion And AUTOMATIC1111

stable-diffusion-photoshop-outpainting-example-automatic1111
After installing the plugin, open the Plugins tab and Plugins panel to access the Auto-Photoshop-SD plugin.

You can also do outpainting in Adobe Photoshop with a Stable Diffusion plugin. Scroll down the Github page for installation instructions on how to do outpainting straight in Photoshop. When you create a new file in Photoshop, remember to uncheck the ‘Artboards’ option.

You need to have AUTOMATIC1111 WebUI running locally, and you also need to modify the webui-user file to have the following script: COMMANDLINE_ARGS= --api

stable-diffusion-photoshop-script-command
COMMANDLINE_ARGS= --api

With the plugin, you can access all of your downloaded diffusion models straight from Photoshop, do outpainting and inpainting, and even use ControlNet to tweak your AI-generated images any way you want.

Outpainting In Playground AI

playground-ai-outpainting-example
Outpainting in Playground AI. You can choose any pixel size to outpaint.

Playground AI offers 1,000 credits per day, and the cool thing about Playground AI is that you can start the outpainting process with any model they currently have on offer.

After you’ve generated an image you like, click the ‘Actions’ button and select Edit in Canvas to start the outpainting process.

Select the ‘Generate image’ option from the canvas area to enable the outpainting.

playground-ai-outpainting-example-image
I selected both images (the original and the outpaint) (image on the left), combined them into ‘image-to-image,’ and generated a new image (image on the right) with ‘Image strength’ set to 30.
playground-ai-outpainting-example-image-final-result
The ‘final’ result of trying out the Playground AI’s outpaint and canvas capabilities.

The randomness of AI affects how poor or good results you get with outpainting. It seems Playground AI is using Stable Diffusion V1.5 or V2.1 for outpainting. The best results can be achieved by combining the generated outpaint (expanded area) with the original image and then using the image-to-image method to generate a new image.

While the mentioned method won’t keep the original image intact, it will help you expand the subject and elements in both images to new pixel sizes and keep the image consistent. Something the current outpainting capabilities lack in AI art generators is making the overall image look consistent with the outpainted area with the original image.

playground-ai-outpainting-example-image-final-result-example-2
The above image has the outpainting result on the right side and the original image on the left. By combining them and using the same text prompt, you can combine the results while keeping the elements from both images.

The downside with Playground AI’s canvas is that you can’t select a specific diffusion model for the outpainting process. For example, if you’ve created an image using the ‘Retro anime’ diffusion model (or a filter) and tried to do outpainting, it will likely fail (more or less) due to the model incompatibility used in the outpainting process.

playground-ai-retro-anime-example
Playground AI’s Retro Anime style on the left side, and the outpainting results (using the same text prompt, including the retro anime style prompt) on the right.

Outpainting In Getimg

outpainting-getimg-demo-image
Outpainting in Getimg. On the left is the original generated image, and on the right is the image extended by outpainting.

While Getimg offers outpainting capabilities, getting good results that blend well with the originally generated image can be time-consuming and frustrating. Getimg offers inpainting (Eraser tool) that enables you to correct parts of the image, but still, the exact vision you have in your mind might require a lot of credits to come to fruition.

Getimg’s AI editor remembers your last edits, so you can continue where you left off. The downside is that Getimg’s AI editor only offers (currently) Stable Diffusion V1.5 and Realistic Vision (most likely V1.0) diffusion models to be used with outpainting. Another downside compared to many other AI art generators, it only offers 512px x 512px editing area.

Feature image credits.

Search
artist-profile-picture-avatar

Okuha

Digital Artist

I’m a digital artist who is passionate about anime and manga art. My true artist journey pretty much started with CTRL+Z. When I experienced that and the limitless color choices and the number of tools I could use with art software, I was sold. Drawing digital anime art is the thing that makes me happy among eating cheeseburgers in between veggie meals.

More Posts

Thank You!

Thank you for visiting the page! If you want to build your next creative product business, I suggest you check out Kittl!