Experiments with AI art workflows

Hello, journal. For the last few days, I’ve been playing around with taking AI-generated images from Midjourney and running them through a bit of workflow using Photoshop and Illustrator, resulting in a full vector image. The goal here is to be able to manipulate better what Midjourney creates.

My biggest hurdle at the moment is my lack of skill with Illustrator. I don’t know how to use it well, let alone optimize things. I can correct colors and simple shapes, but making the paths more efficient and more complicated tweaking is beyond me. Still, I think something is interesting here.

I’ve also tried hand tracing, which works much better. It’s a bit time-consuming, but it’s fun and engaging. The results are much better, as you might expect, but I’ve run up against barriers primarily because of my lack of knowledge of the tools. In this case, I’ve been using Figma because Illustrator feels difficult.

You can see some progress here, the Midjourney image on the right and my vector drawing in Figma on the left. I quite like how it’s shaping up.

I think there might be some workflow that uses a combination of these. Once I finish working on this current conversion, I think I’ll dive back into Midjourney. I think there is a lot I could do to work the prompts to get something cleaner, to begin with. Then I could use a combination of Image Trace in Illustrator and hand tweaks to get something nice. I might even start experimenting with Procreate to add some hand-drawn flourishes. We’ll see.