The time has come to say goodbye to iStock, Getty, Unsplash, and the many stock photo sites that overcharge and underdeliver on photo authenticity.

Whether you're building a new website, promo material, or presentation deck, images are critical to liven up content and allow your audience to better connect with it.

While AI image generation continues to improve, we're now at a time when you can reasonably replace many types of stock images with AI-generated images. In particular, using Midjourney.

Here's how to do it:

1. Determine your first prompt & benchmark on Unsplash

Let's assume we're trying to create an image for a website that shows a person working in a coffee shop on a laptop, with the first prompt being:

a hipster working in a trendy coffee shop on a laptop

Just to get things flowing in a familiar environment, head over to unsplash.com and see what you can find with a quick initial search.

You'll likely get some high-quality imagery from Unsplash, albeit quite unrelated to the prompt, and then dozens of "stock-y" images advertised to you from iStock.

The free section is not relevant enough for what we're looking for, and the paid looks way too overly staged, so let's go over to Midjourney!

2. First Prompt in Midjourney

Now let's roll over to the leading generative AI image platform, Midjourney, and try it out. If you haven't created an account yet on Midjourney, you can see how to do that here. Sadly, you still need to do it through Discord, and only through a paid account, but their web version is coming soon!

Before starting, let's take a quick peek at our image settings in Midjourney (in Discord).

To access and adjust these features, enter /settings in the chat prompt. We can play with these more later, but I generally start with "Stylize med," and ensure Remix mode is ON.

Let's drop the following prompt into Midjourney. I added --ar 3:2 so that we get a 3:2 aspect ratio.

a hipster working in a trendy coffee shop on a laptop --ar 3:2

Hipster Coffee Shop: Generation #1

Not bad! All of them are better than anything I saw on Unsplash, but let's see how much further we can push it. I will generally do 3 to 4 straight re-dos of the same prompt to see what happens.

You can do this by tapping the white and blue arrow button under the images.

Hipster Coffee Shop: Generation #2

I like the different styles we got here, particularly the comic book style variation it generated! Let's try one more and then go back for some refinements.

Hipster Coffee Shop: Generation #3

A bit more of the same, although I like how we're seeing the street outside a little bit. From here, I'll pick 1 or 2 that I like and then push on them, depending on what I'm using the image for.

3. Remix the prompt

For this exercise, I'm going to pick two images and play with them a bit.

I like the composition of this one, but I think we can do better. First, I tapped the U3 in the grid beneath to upscale that image and now I get all kinds of fun options. I'm also going to play with the comic book one, as I think there is some fun to be had there.

There is a lot of detail you could get into here editing the prompt, but first I like to see what I can get by straight "varying" subtle and strong. Now I'm going to get 8 more versions of each of these images to play with!

Sweet! Getting some nice images here now. Lots to play with and experiment with. For this post, I'm intrigued by pushing the comic book version a bit further so let's zero in there.

4. Edit the prompt and remix again

I love this one, and keen to see what we can do with it.

Let's "Vary" Subtle and Strong again, but this time let's adjust the prompt to make it a photograph, instead of a comic/sketch drawing.

a photograph of a hipster working in a trendy coffee shop on a laptop --ar 3:2

Once again, this will now give us 8 more images to look at and play with.

Here's where you can sometimes get some weird stuff, which we did here. But, I think we have one we can focus on one more time.

We went backward on some of these, but I like this one (below). That said, there is a weird "double laptop" thing happening on the table. It's always good practice to keep your eyes out for unnatural elements dropping into the frame. The models are good enough to avoid this, you just need to keep pushing.

Let's try and REMIX this one again and see if we can get rid of the rogue laptop.

I think I've found one that is pretty close. There is something kind of strange happening with his hand, but I it's close enough for now. It is far better than anything we could have purchased or found on a free site!

5. Upscale and export

The final step is to upscale the image, which again, I normally do both a subtle and creative upscale to see what I can get at the final stage. Finally, click on the 'Web' button to view it on the Midjourney website and download it.

While this can take some time to get it right, it is often more because of the specificity that is possible and what you can change, versus just accepting some random stock image.

This post is just scratching the surface of what is possible, but I highly recommend every designer, marketer, and product manager give this a go and hopefully say goodbye to many of the stock sites for good!

How to create "stock images" using AI

Time to say goodbye to stock image sites. Learn how to easily create exactly the photo you need for your project.

The all-in-one writing platform.

Write, publish everywhere, see what works, and become a better writer - all in one place.