Streamline Image Creation With AI

Prepare images for your published content.

Streamline Image Creation With AI

For more than a year, I've been using Midjourney to create images for my Substack, newsletter, Medium, and blog.

Pre-Midjourney, I used stock photos from Unsplash. I really wanted unique images for my writing, but I didn't have a way to create them. Filters were the best I could do, giving the photos a monochrome vibe.

Important: I fully believe artists and writers should be fairly compensated for their work. Midjourney is not a substitute for an artist and companies that can afford illustrators end up with artwork that is much better than what I can produce with Midjourney.

I'm in a spot with my work where my options are "use stock photos" or "try to generate something a little more unique using AI." Between those two choices, AI is better suited for what I'm trying to do.

This isn't a tutorial about "how to use Midjourney." For that, you can check out Midjourney's Help Docs. This tutorial assumes you know how to use Midjourey and follows my process for streamlining AI image creation.

I'd get stuck at the Midjourney prompt stage. I know how to prompt Midjourney, but I couldn't envision the output I wanted.

I've gotten past "image block" (like writer's block, but for images) with a little help from Unsplash.

Step 1: Find an image on Unsplash

I start my process in the same way I did pre-Midjourney: in Unsplash. I'll type in a few words or a phrase and start browsing photos.

Search in Unsplash

Once I've found an image, I'll download it.

Download an image from Unsplash

Step 2: Use the image as inspiration for Midjourney

To create new images in Midjourney, you use the /imagine command.

In my case, I don't want to start from scratch. I want to use my Unsplash image as inspiration. So I prompt Midjourney to describe the Unsplash image. The /describe command will return what Midjourney "sees" in the Unsplash image. I can then use this to create my own prompt.

The /describe command has two options: image and link. I select image.

The /describe command in Midjourney

Midjourney will prompt me to either drag-and-drop or click to upload my image file.

Image upload in Midjourney

I'll browse to my downloaded Unsplash image and select it.

File explorer of Downloads folder

My selected Unsplash photo will be displayed in Midjourney. Hit "Enter" on the keyboard to send the /describe command.

Midjourney will return four descriptions of the image. Usually, the first or second is closest to what I want.

I'll highlight the text, right-click, and select "Copy."

Copy text from Midjourney prompts

Then, I'll run the /imagine command. I paste the text I'd copied from the /describe result.

This is where I edit the prompt. In this example, Midjourney came back with "photography in the style of Unsplash" as part of the description. I don't want that in my new prompt! So I'll take that text out.

I also use different styles for different publications (Substack = graphic novel, Medium = pop art, blog = abstract watercolor illustration). I add the style to the beginning of my prompt.

Midjourney /imagine command with pasted text

Midjourney will return four image results. I'll select the one I want and upscale it.

Midjourney image results

Then I'll right-click on the image an Save it.

Dsbran image in Midjourney

Step 3: Shrink the image with Squoosh

Midjourney's file sizes are quite large – much bigger than I want on my website. It can affect the performance of my page. So I need to shrink the image.

To do this, I use Squoosh (which is free).

Click on the "Drop OR Paste" button in the middle of the screen. You'll browse to your image and select it.

Home screen of Squoosh.app

Once your image is uploaded, you'll click on the down arrow button in the lower right corner of the screen. Squoosh will default shrink your image by over 90%, but you can change this by adjusting the Quality.

Download an image in Squoosh

That's it! You can click on the X in the upper right corner of the screen to close the image (and upload another, if needed).

Close Squoosh app

At this point, I'll usually rename my image to something like sitename-short-description-of-post.

Step 4: Generate alt text with Writer

Next, I need to generate Alt Text for my image. I do this in Writer (I'm on a paid plan).

In Writer's apps, I select Image analyzer.

Image analyzer app in Writer

I browse to upload my image.

Upload an image in Writer's Image Analyzer

In the prompt, I type in "generate alt text" and click the Generate text button.

The output is usually one or two sentences. I'll edit it to something shorter.

Output of alt text from Writer

Step 5: Add the image to my content library in Airtable

Finally, I'll upload the image to my Google Drive.

My Drive folder is synced to my Airtable Content Library, so I'll see the image added as a new record.

I'll link the image to the appropriate article, once it's published.

Images in Airtable synced from Google Drive

More resources:

I created the screenshots in this tutorial using Scribe – it's one of my favorite tools. You can check out Scribe here [affiliate link].