3 Lesser-Known Features of Midjourney's Zoom Out

PLUS: latest updates from Midjourney, Stability AI, and Adobe

In this newsletter, read about:

  • 🕵️‍♀️ Zoom Out in Midjourney

  • 🗞 News and Top Reads

  • 📌 AI Art Tutorial: Midjourney Upgrade

  • 🎨 Featured Artist: Josh Larson

  • 🖼 AI-Assisted Artwork of the Week

🕵️‍♀️ Zoom Out in Midjourney

The top news from the last week is definitely Midjourney releasing v5.2. This release includes many exciting updates, including:

  • improved aesthetics and sharper images

  • higher level of stylization (--s 1000 in v5.1 is equivalent to --s 200 in v5.2)

  • high variation mode (turned on by default, but can be adjusted)

  • new /shorten command to understand the key parts of your prompt

  • and finally, the new cool Zoom Out feature.

In this post, I’ll focus on the highly anticipated Zoom Out feature, its lesser-known capabilities, and also some limitations.

Intro to Zoom Out

Basically, the Zoom Out feature works pretty straightforwardly. After upscaling your favorite image from an image grid, you see 4 buttons related to Zoom Out:

  • Zoom Out 2x

  • Zoom Out 1.5x

  • Make Square

  • Custom Zoom

With the first two options, you kind of pull the camera out and fill in new details on all sides of the images.

a portrait of a black man in a surrealistic style --ar 4:5 --s 750

Original + Zoom Out 2x

In this case, the aspect ratio and resolution of your image stay the same, but you get more environment around your subject.

This is how Zoom works by default, but the latest Midjourney upgrade gives you some additional tools that not everybody is aware of.

Advanced Features

1. Changing aspect ratio

If it happens that you generated a vertical or horizontal image, and you want a square one instead, you can simply choose the Make Square button, and Midjourney will add details on two sides and transform a non-square image into a square one.

a person with enormous wings made of butterflies --ar 4:5

Original + Make Square

In the example above, the original image was generated with cut-off wings, and I leveraged the Make Square feature to get full wings on the image.

If a square image is not what you are looking for, you can use Custom Zoom to get any aspect ratio you want. After you hit this button, you’ll see the text box with your original prompt and --zoom factor set to the value of 2, by default. Now, you are free to change the --ar parameter to any value you prefer.

If you only need the aspect ratio changed, without additional zooming, set the --zoom parameter to the value of 1. If you prefer a little bit of zooming, adjust the --zoom parameter accordingly (e.g., --zoom 1.1. or --zoom 1.2).

a portrait of a black man in a surrealistic style --ar 4:5 --s 750

Original + Custom Zoom (--ar 16:9 --zoom 1.2)

2. Changing the environment

In fact, Custom Zoom gives you much more freedom than just adjusting the aspect ratio and zoom value. You can completely change the environment by modifying your text prompt. The results might be strange, interesting, or brilliant. The experiments are usually fun! 😀 

a man transforming into a statue, light colors, surrealism --ar 16:9 --chaos 50

Original image

city park, a man transforming into a statue, light colors, surrealism --chaos 50 --ar 16:9 --zoom 2

In this example, I simply added “city park” at the beginning of the original prompt, but you can actually rewrite the entire prompt if you want. You’ll still get the resemblance to the original image in the center of your new image, but the environment is likely to be less connected (depending on the new prompt).

city park, light colors, surrealism --chaos 50 --ar 16:9 --zoom 2

3. Applying Zoom Out to earlier images

I have noticed many people wondering whether it’s possible to apply this new Zoom Out feature to the images they generated with an earlier Midjourney version. And the answer is positive! You can zoom out images you generated previously with v5 or v5.1.

But first, you’ll need to find the Job ID of your initial generation. There are two options:

  • In your Midjourney gallery, find the image and choose … → Copy → Job ID.

  • In your Discord, find the message with the upscaled image and react with an envelope emoji. You’ll get a new message with the Job ID and seed value, like the one below.

Then, type the command, like this but with your Job ID:

/show job_id b73b962f-8b11-418c-9e48-7f7756e490af

And you’ll get all the new v5.2 features available for your old image:

Original + Zoom Out 2x

Limitations

Zoom Out feature offers a new level of flexibility for Midjourney creators, and I think it performs on par with Photoshop’s Generative Fill feature while allowing commercial use for generated images. However, we should also acknowledge some characteristics and limitations of this tool that might require improvement in future releases:

  • Zoom Out doesn’t improve the quality of output images, implying that zoomed-out images have about the same resolution as the original ones.

  • A lot of users noticed that v5.2 tends to generate darker and darker environments with every new zooming iteration. If you encounter this problem, try Custom Zoom and adjust the prompt with the request for light / bright colors.

  • Midjourney doesn’t “understand” the image. When you request Zoom out without additional prompts, the algorithm simply tries to guess the best option based on the edges of the original image and your original text prompt. Hence, the generated environment will often seem strange and irrelevant, especially when you experiment with photorealism. Still, you can get some stunningly creative and original environments for artistic images.

Before working with any new tool, it’s important to understand both its strengths and limitations. So, I hope this brief guide will help you get the best from this handy Zoom Out feature in Midjourney.

Happy prompting!

🗞 News and Top Reads

  • Midjourney released version 5.2 with new improved aesthetics, higher stylization, and the popular Zoom Out feature.

  • Stability AI revealed SDXL 0.9, their latest diffusion model for realistic image generation.

    • The model can generate hyper-realistic creations for films, television, music, and instructional videos.

    • The SDXL series also offers a range of functionalities text-to-image prompting. These include image-to-image prompting (inputting one image to get variations of that image), inpainting (reconstructing missing parts of an image), and outpainting (constructing a seamless extension of an existing image).

  • Adobe guarantees customer protection from third-party IP claims regarding Firefly-generated outputs.

    • This move aims to attract big enterprise clients who prioritize the commercial viability of images and typically favor human-created stock images over AI-generated content, which currently resides in a legal gray zone.

    • To enable such protection, they limited their training dataset to either Adobe Stock or open-source images.

    • Note that images created with Adobe Firefly Beta are not for commercial use.

📌 AI Art Tutorial: Midjourney Upgrade

In this video, Matt Wolfe discusses the cool new abilities we have with the MidJourney 5.2 update. He also includes tutorials on how to get multiple characters into a single scene, how to create a consistent character in different environments, as well as how to do the "infinite zoom" effect.

🎨 Featured Artist: Josh Larson

Josh Larson, a seasoned game developer and digital artist, boasts over 20 years of industry expertise. Notably, he is the founder of Numinous Games studio, renowned for its emotionally-driven storytelling masterpiece, That Dragon, Cancer. In June 2022, Josh joined Midjourney as an esteemed private tester, contributing his insights to the platform's development. Additionally, he had the honor of being the inaugural creator to receive a comprehensive 12-page feature in the first issue of Midjourney Magazine.

🖼 AI-Assisted Artwork of the Week

Share Kiki and Mozart

If you like this newsletter and know somebody who might also like it, feel free to share this newsletter. Let’s have more people learn about AI art!

If you have been forwarded this email and you like it, please subscribe below. And welcome to the world of AI art!

Join the conversation

or to participate.