You've probably heard about ChatGPT and that artificial intelligence software is becoming more accessible to the general public. Using AI image generation for design is becoming more common as well, and is a fun tool to play around with.
We'll walk through designing a rug hooking project using AI from start to finish.
DALL-E is an AI image generation software. There are others available, and google is working on a version that isn't yet available. An account is needed to access the software, and credits are provided free and renewed on a monthly basis, or credits can be purchased.
This software works by the user inputting text (specific text prompts work best), and 4 images are generated. These images are generated by the software, therefore there is no limit to what can be imagined/created.
For example, the text prompt "expressionism painting of a lily" resulted in the following four images:
A more general query "expressionism painting of a flower" resulted in the following:
If one of the images generated is appealing, you can request for variations (using another credit). Here's the variations proposed for the purple flower above:
The same query can be submitted multiple times and will produce four different images each time. This means that two users can input exactly the same query and see different images. You'll notice that all of the generated images are square, and this is not changeable at this point, but the image can be edited within the software.
Being specific in the query is helpful. For example, stating the type of image you're looking for (photograph, realistic, 3D, painting, outline, etc.), as well as color, background, or other aspects will help generate a better image.
The query: "an abstract oil painting of a flower" resulted in the following:
The fourth image here caught my eye. Images can be downloaded directly, in PNG format.
Now is a good time to explain who owns the images generated in this way. Currently, DALL-E indicates that "...you own the images you create with DALL·E, including the right to reprint, sell, and merchandise – regardless of whether an image was generated through a free or paid credit." They do request that credit is given to DALL-E, in the following manner: "This image was created with the assistance of DALL·E 2" or "This image was generated with the assistance of AI."
Now to turn this image into a rug hooking project!
First, the image was printed, and the flower itself was traced using Red Dot paper and transferred to monks cloth (more on transferring patterns using red dot paper here). 
Loopy Wool sells cut strips, and therefore there are a lot of wool scraps in the studio. Often these are tiny (a #2 or smaller), the rind of the wool, or misshaped, and for this project I opted to use only pieces of scraps. Sometimes I needed to snip a piece in two, but generally I was able to grab from this mess and find the right color.
Keeping the image open on the screen to follow along, the hooking commenced!
For the background I didn't bother drawing any lines and just freely followed the image.
When most of the hooking was finished I went back to add accent wool pieces. If the hooking isn't too tightly packed (more on that here), it can be easy to squeeze in extra pieces to add highlights/differentiation. In some cases the strip was cut using scissors rather than a cutter to get very small strips.
Even when it looks like there might not be room to add a strip, it can usually be done!
The pile of scraps didn't seem to diminish over the project!
And here's the finished piece. Designed by Chris Waddy with assistance of AI, and hooked by Chris Waddy, 2023.
Here's another example of image generation and the final project.
Query: "Outline of abstract flowers"
This piece was punch needled using variegated wool.
