• Kiki and Mozart
  • Posts
  • Midjourney's Inpainting Feature: Take Control of Your Creations

Midjourney's Inpainting Feature: Take Control of Your Creations

Learn how to edit subjects and adjust colors with Inpainting

In this newsletter, read about:

  • 🕵️‍♀️ New Levels of Output Control With Inpainting

  • 🗞 News and Top Reads

  • 📌 AI Art Tutorial: Midjourney Inpainting

  • 🎨 Featured Artist: AI.S.A.M

  • 🖼 AI-Assisted Artwork of the Week

  • 🤓 A Comprehensive Midjourney Guide

🕵️‍♀️ New Levels of Output Control With Inpainting

Midjourney has finally released the long-awaited Inpainting feature. Now you can select a specific region in your output image and guide Midjourney to re-roll this region with a new prompt.

Let’s start with a brief introduction to Inpainting and then we’ll have a few examples to see how this feature gives a whole new level of control with regard to colors and other characteristics of your images.

How to Use

To edit a specific region of your image, follow these steps:

  • Go to /settings and make sure that you have Remix mode switched on. With this mode on, you’ll get a text box in your editor that will allow you to modify the prompt for the selected region.

  • Upscale an image that you want to edit.

  • Click on Vary (Region).

  • In a new Editor window that will pop up, select the region to edit.

    • Rectangular and Lasso tools are available for region selection.

    • It’s possible to choose multiple regions.

  • Within the same window, edit the prompt so that it only includes what you want to get in the selected region.

  • Press an arrow next to the prompt, close the Editor window, and wait for the re-rolled results.

A few things to note when using Inpainting in Midjourney:

  • The feature is still experimental and may output unexpected results.

  • It works best on large regions of the image (20% to 50% of the image).

  • The output will be more reliable if your new prompt is a good match to the image. For example, it would be easier to have a woman holding a purse in her hand than a crocodile.

On the whole, it can be quite fun to play with this feature. Let’s test it with a few tasks that used to be extremely challenging before the introduction of Inpainting.

Edit Subjects

It is usually very difficult to have Midjourney depict two subjects with opposite emotions in one image. The first emotion you mention in the prompt will be usually applied to all subjects. Can we fix this with Inpainting?

a sad woman and a happy man, 1960s style --ar 16:9

As expected, we got both subjects sad in the initial output image. Now, let’s have a man re-rolled with a modified prompt: “a happy man, 1960s style --ar 16:9”.

a happy man, 1960s style --ar 16:9

Great result from the first attempt! Now it’s much easier to get multiple subjects with distinct characteristics in one image.

Control Colors

Another challenge we have with Midjourney is controlling the colors in the output images, especially if we have a very specific request about distinct image parts.

For example, it turned out to be quite challenging to get a woman in a blue skirt, yellow top, and a red jacket with all the output images switching colors one way or another.

a photo of a black woman in blue skirt, yellow top, and red jacket walking confidently down the city street --ar 4:5

This image has a jacket of the right color. Now, we’ll try to adjust other colors using Inpainting. First, let’s change the skirt to the blue one.

a blue skirt --ar 4:5

Then, let’s switch the top to the yellow one.

a yellow top --ar 4:5

And here we arrived at what was requested initially. As you see, with Inpainting it becomes much easier to control colors in your images.

To Sum Up

Here I covered just a few use cases, but it’s already obvious that Inpainting offers a whole new level of control and flexibility.

However, note that it’s not a magic fix. You’ll still notice Midjourney fighting your intents and giving some frustrating results. From my experience, Inpainting also tends to output some strange and unexpected results, as if you have a chaos parameter set high.

Still, it’s fun to play with and I love the new opportunities this tool provides. Looking forward to what Midjourney has yet to offer.

Happy prompting!

🗞 News and Top Reads

  • Microsoft Designer, a free AI-powered design tool from Microsoft is now available to Edge users in the US.

    • You can access Designer from the Edge sidebar, so you don't have to open a separate tab or program.

    • Simply enter a prompt, and Designer's AI will generate a grid of suggested designs, each with text and a photo. You can then customize any of these designs to your liking, such as swapping out the photo, adding an illustration, or changing the font.

  • Snapchat is expanding its generative AI capabilities, and is currently working on the two new features:

    • Dreams: This feature will allow users to place their selfies in fantastical places and scenarios. The AI will generate new images based on the user's selfie and their input on the desired background and scenario.

    • Dreams with Friends: This feature will allow users to give their friends permission to generate AI-generated images that feature both the user and their friend.

  • Adobe has released its Adobe Express with Firefly web app from beta, making its AI-powered image tools available online. A mobile version is coming soon.

  • Alias Technologies has introduced BeFake, a social media app for digital self-expression.

    • The platform encourages users to express their authenticity through AI-generated images.

  • AI-generated art cannot be copyrighted in the US.

    • In a recent ruling, US District Court Judge Beryl A. Howell affirmed the US Copyright Office's decision to deny Stephen Thaler a copyright for his AI-generated image.

    • Interestingly, the image in question was created using Thaler's Creativity Machine algorithm, but the judge found that it was still a work "absent any guiding human hand."

    • We can only hope that regulations will catch up with new technologies sooner rather than later.

📌 AI Art Tutorial: Midjourney Inpainting Tutorial

For those who favor visual learning, Julian van Dieken, an esteemed AI artist, offers a concise 1-minute Inpainting tutorial. The video demonstrates the use of the Rectangular and Lasso tools for selecting areas to edit, as well as how to adjust your prompt – especially if you wish to completely remove an element from your image.

🎨 Featured Artist: AI.S.A.M

AI.S.A.M or Sam Finn is a London-based AI photographer, concept artist, and creative director. His work draws on themes of decay, loss, and the supernatural. He uses AI to challenge what is real and artificial, to push the boundaries of traditional artistic practice, and challenge preconceived notions of what is possible with technology.

🖼 AI-Assisted Artwork of the Week

🤓 A Comprehensive Midjourney Guide

To get a link to a comprehensive Midjourney guide, please subscribe to this newsletter. The guide is a dynamic document, which I intend to keep up-to-date with the latest Midjourney updates.

Share Kiki and Mozart

If you enjoy this newsletter and know someone who might also appreciate it, please feel free to share it with them. Let's spread the word about AI art and introduce more people to this fascinating field!

Join the conversation

or to participate.