Back in early 2024, all I was hearing about was generative AI and how it would change the world. There has been so much debate about whether the changes would be good, bad, beneficial, destructive, or all of the above. This debate has been quite intense in the art world, though maybe not so heavily discussed in the quilting world. Because of my day job as a transcriber, I have been well aware that technology will be replacing me at any point, but I had never really considered how it could affect my creative endeavors until I started seeing everyone’s experiments with image generators. I decided to find out for myself just how afraid or excited I should be about using this cutting-edge technology in combination with a millennia-old craft and art form.

The Brooklyn Quilters Guild had an outdoor fence show in the autumn of 2024 that was titled The Stories We Tell: Folklore, Fantasy, and Folly! I almost immediately had an image come to mind when I started thinking about quilt ideas for this show. I imagined a person staring into a cracked mirror. The reflection’s arm is reaching through the mirror holding some sort of object representing power. Kind of a “Snow White/Through the Looking Glass/Garden of Eden” type deal. The problem with this idea was that I do not have the artistic skills necessary to draw the kind of image I wanted to create, so what a perfect time to dip my toe into the world of generative AI, right?

I decided the best way to transfer what I was imagining into an actual quilt would be to use the raw-edge appliqué technique. Interestingly, this technique already takes advantage of quite a few forms of technology that weren’t available to quilters of yore. You generally use double-sided fusible interfacing if you’re creating a raw-edge appliqué design. Many folks use photographs as the foundation of their quilt images. Then you can use a printer or a projector to trace the portions of the photograph into the different sections you’ll eventually appliqué together. So why not add a little generative AI into the mix? It’s actually not that big of a leap when you really think about it.

Next, I had to choose a program to create the image. After a relatively quick Google search, I opted for Microsoft Image Creator (which is now called Microsoft Designer). I chose this program because it was free and offered unlimited prompts (I’m not sure if that’s still the case today).

I entered my first prompt:

gilt frame antique mirror with shattered glass reflecting the image of a person and a hand reaching out holding an apple

I proceeded to tweak the prompt 12 more times. I’ve posted all of the resulting images below.

My final prompt was:

in the style of an oil painting, a nonbinary person staring into a shattered gilt frame antique mirror set at an angle. A completely separate hand holding an apple coming out of the mirror

These were the three images that came closest to what I was picturing in my mind.

While none of these images were exactly what I wanted, I decided the one on the far right was close enough and I could make changes as I was putting it all together. You may also notice that all of the figures are holding an apple. I had no plans to use an apple as my symbol of power, but “flame-like blob of power” was a little too abstract for the image generator.

Once I chose my image, I used Blockposters.com to enlarge the image to the size I wanted and then printed it out on my black-and-white printer. And because I was using the raw-edge appliqué technique, I needed to mirror the image and print out that version as well since everything you trace on the back of your fused fabric will reverse once you start putting all the pieces together. I suppose I didn’t have to print out the non-mirror image, but I found it very useful to have the full-sized image on hand as a reference when I started putting all of my many bits of fabric together.

I won’t go into all the nitty-gritty details of how one creates a raw-edge appliqué quilt (there are tons of online videos/workshops/classes if you’re interested), but I will say this was by far the most ambitious raw-edge appliqué quilt I’ve ever attempted and it was exhausting. I don’t know how all y’all do it! Respect.

This is how my quilt turned out.

You can see that it looks fairly different from the final AI-generated image I chose. I intentionally tweaked the design as I was cutting the little bits of fabric out and deciding what details from the original image could be ignored and what details needed to be changed. While I can genuinely say the final product is my own design, I never could have gotten to this point if I hadn’t used the AI-powered image generator.

So after all of this, how do I feel about using AI in my making process? I’d say I’m pretty jazzed about it. For someone who has limited drawing skills, I am looking forward to experimenting further with AI image generators used as a tool to help me create more projects like this one. From an ethics point of view, I do believe one should be transparent about how AI was used in the process of creating a work of art, but I believe it can be used as a transformative technology to help create fascinating results.

Published by Andrew Ve Hansen

I live in Brooklyn with my husband. I'm obsessed with all things quilting! Some of my other interests include taking advantage of all of the culinary delights this city has to offer, hanging out with my friends, board games and tabletop RPGs, reading, watching movies and tons of TV, crafting, and going to the theater (especially musicals).

Join the Conversation

  1. Laura Kate's avatar
  2. Andrew Ve Hansen's avatar
  3. Helen's avatar
  4. Unknown's avatar

5 Comments

  1. Your quilt is amazing.

    I am fascinated by AI and curious about inherent biases from programming. I notice that your initial prompt was for a person but all the images generated are men and almost all of them are white. Your final prompt was for a non-binary person and they still all look very male (that might be my bias there about what a non-binary person looks like).

    1. I completely agree, Helen! Ideally, I wanted both the figure outside the mirror and the reflection to be something that anyone looking at the quilt could possibly relate to no matter how they identify as far as gender, race, etc. But, yeah, I couldn’t figure out how to get the AI to do exactly what I wanted. I’m sure part of that is programming bias in the image generator, but I also only entered a total of 13 prompts. And I generated those images a year ago, so maybe things have changed since AI seems to be adapting very, very quickly.

Leave a comment

Leave a reply to Andrew Ve Hansen Cancel reply