I don’t just write a prompt and have an AI directly generate my image. However, I use AI tools to optimize my workflow.
Code generation
I’ve implemented my own software pipeline to efficiently realize my ideas, first into a composition on my computer and then brought to the physical world with computer-controlled machines (laser, cutting, milling).
My process and therefore my software pipeline is fluid, constantly evolving with each new project. This is for me meta-creativity, creativity that spawns possibilities for other creativity.
The pipeline was coded with the assistance of an AI tool. The architecture and structural engineering were done by myself. AI did 40% of the low-level coding.
Text
As someone who is as dyslexic as a rabbit, I rely on large language models to help format texts, such as this one.
Segmentation
One of the 17 modules within the pipeline is for image decomposition, a.k.a. segmentation. I’ve built this module around an AI model of Meta called “Segment Anything.” For example, in my piece “Saint Johnnyboy,” the model assisted to find the outlines of the cathedral.
Drawing
This is the hard part, the part where the discussion is.
When I need a drawing, I can ask an AI for a proposal. These drawings serve as a starting point, which I then modify and incorporate as one of the components into my compositions. I’ve used this method in my early work: “Brainzipper,” “Hansie Dansie” and in my logo.
I struggle with the stigma AI image generation has. Do I ignore the stigma, fight the stigma, or be a sheep and just not use AI drawings to prevent any discussion? At this moment in time, I choose to be a sheep and put it on a hiatus.