Drawing your IFS Parts with AI Image Generation (using Midjourney)

This article presumes you already know something about IFS therapy and have a few hours to learn how to generate AI art. I provide links to useful resources, but the main purpose of the article is to show you the end result and a good set of tools to get there, not to provide step-by-step instructions. Here we go!

It occurred to me after a recent IFS session with my therapist that it would be fun to create AI-generated images of my parts. I know I’ve really identified a part when I can feel it in my body at a specific location, AND an image pops into my head. Thereafter, I can use the image to recall and speak with the part instantly. Having it on paper would make it even more real, and maybe even help my therapist help me.

I had already created a map of my parts, using post-it notes on a large piece of paper. This was a suggestion from the excellent “Self-Therapy Workbook” by Bonnie J. Weiss. If you’re into IFS, get the book.

Here is the result:

The parts map is useful because it helps me investigate the relationships between my parts, but it sure ain’t pretty.

So I thought, hey, I use ChatGPT and Pi and even installed an Alpaca LLM on my MacBook (ok, a friend gave me detailed instructions), how hard could it be to use an AI image generator to create images of my parts? Quick answer: it took me about 4 hours, but now I feel like a super-hero! I look like one too, here’s the image I created of my Self:

That, um, kind of looks like me! I liked it so much that I’m using it now as my profile pic in lots of places.

Here’s how I did it. First I tried Microsoft’s Bing AI image tool, which as billed, is super easy to use, and I used to create Bizzy at the beginning of this article. But then I typed in this prompt: “5 year old boy with dark hair, sitting on the ground.” and got this:

Seriously? So I decided to bit the bullet and learn how to use Midjourney (it’s one word, not two, not hyphenated). It wasn’t a great start, a less attractive home page would be hard to imagine. First I learned that you have to know how to use a Discord server to use Midjourney. It’s not a website. Hopefully you have a video game addiction or some other reason to have learned Discord, otherwise it’s another learning curve. I also paid $9 for a 1 month membership because Midjourney isn’t doing free betas any more. And learning the arcane interface of Midjourney itself took some more reading — what ever happened to designing intuitive buttons? The Midjourney userguide on the site is helpful, but I mostly bounced around approachableai.com for things like finding the best settings. And this site for learning how to create AI images of myself.

And here’s what happened when I asked Midjourney to create my child part picture. After this, I only used Midjourney. Here’s the prompt that after a few rounds of trial and error eventually matched image in my head, “2 year old male child, sitting on ground, staring at camera, east indian, fair skin, blue and white horizontal striped golf shirt. Baby like features big eyes, big head.”

Bam! Off to the races. Then I created all my most important parts, put them into a Powerpoint slide and printed it off.

Wow, did that feel good. Your turn!