Exploring The Potential Of Midjourney For Image Localization

Image generated with Midjourney

Can AI art generators rise to the challenge and help localize images for local audiences?

With all the hype around AI developments, I jumped on the bandwagon a few months ago to explore the intricacies of text-to-image conversion.

For various reasons, the main one being an excessive engagement in the Web3 community, my go-to tool was Midjourney.

Cutting To The Chase

Can Midjourney & Co. be useful for image localization?

The short answer sounds: No (for now).

That is if you consider “image localization” in the traditional terms of using an original image to tweak the colors, texts, cultural items, or ethnicities to suit the local needs.

But if you adopt a more flexible (creative?) approach, my answer is: Absolutely yes.

And by flexible approach, I mean abandoning the idea that image localization is limited to adopting selected items to the target market’s needs.

In this approach, image localization means creating a new graphic from scratch to make sure it resonates with your local audience.

And that’s where Midjourney may come in handy.

Let me show you a few examples of imaginary use cases that I created to test the power of Midjourney.

Localization Case 1

An international chain of coffee shops localizes its website for users in Morocco. The original website presents a white, young American couple drinking takeaway iced coffee on a sunny beach. My target users probably won’t resonate with that picture, so I want to display local consumers on my Moroccan website. Preferably, wearing clothes instead of bikinis. Of course, I could arrange a quick photo shoot and have my perfect image after a while, but I could also turn to Midjourney for help and tweak the image later, if necessary.

Here’s the prompt and the result:

Prompt used: a portrait photo shot with Canon EF 85mm f/1.4, the photo shows two young modern Moroccan women wearing colorful clothes, long dark hair. Both women are in an elegant coffee shop, seating at a table with four cups of coffee. They are smiling, talking, looking at each other, having fun, 8k, hyperdetailed, hyperrealistic, fully lit background

Localization Case 2

A mobile game localizes the content for Chinese users and the new version should include references to the local culture. One of the images will be used for the Chinese New Year. Let’s see how Midjourney could help here.

Here’s the prompt and the result:

Prompt used: a graphic for the Chinese New Year celebration. It should show red Chinese lanterns and the zodiac sign of a rabbit, Salvador Dali style, 4k

Localization Case 3

A mobile security provider expands to France and localizes its digital content for the target users. One of the images used on the website should show a local consumer using a smartphone and local landmarks in the background.

Here’s the prompt and the result:

Prompt used: a body shot of a young French-looking man, walking on a street in Paris, he is looking at a smartphone that he holds in his hand, he is smiling, wearing hipster-style clothes, fully lit background, natural afternoon lighting, 35 mm, ar 4:3

Is Midjourney Fit For Localization Purposes?

Are these creations enough to satisfy the needs of local target audiences?

My judgment leans towards yes.

Depending on your product, image preferences, and target users, you can still tweak the image in another (manual) tool. But even without any modifications, the graphics generated by Midjourney look decent and could be used in digital products to engage with local users.

What About Copyrights?

Can you use AI-generated images freely in your commercial products?

Midjourney states on its website that you can use assets generated in the tool for any purpose, including commercial, as long as you use the paid license. Plus, you’re not allowed to claim ownership or authorship of the images.

Once you pay for the license, you’re free to use your images wherever you want (and yes, I paid for mine).

What About The Text?

There’s one limitation when it comes to using graphics created in Midjourney for localization purposes. The tool struggles with text.

If you want to add text to your design, you will need to seek help elsewhere. Also, if your image localization operations are limited to merely swapping the text in language A for the text in language B, it will probably be quicker to stick to the human-generated art (in Photoshop, Gimp, etc.).

Final Thoughts

AI art generators might be a useful resource when you want to design images for local audiences: you can prompt the tool to create images in any format, quickly create decent results, and use the graphics commercially.

All you need to do is equip yourself with patience — it’s key if you want to learn how to write precise prompts to be able to obtain satisfying results.

What are your thoughts on using AI for image localization?

Have you tried generating culturally specific images with Midjourney?

As a localization consultant, I’m happy to help you make sure your content is fit for your target market.