In my previous work, I've often leaned on custom generative systems, compositions created from code, sampled and transformed with computer vision to reveal structure, flow, and pattern. But with this latest series, I wanted to shift the starting point. Instead of writing the source image into existence with generative code, I used an AI image generator to produce the seed material. This decision wasn't just about changing tools. It was about shifting perspective, starting not with abstraction, but with familiarity.
Each image begins as a seemingly ordinary, AI-generated visual, often scenes from nature, flowers, or textures. From there, I break it down. Using computer vision techniques, I analyze the image and lay a grid over it...
Each image begins as a seemingly ordinary, AI-generated visual, often scenes from nature, flowers, or textures. From there, I break it down. Using computer vision techniques, I analyze the image and lay a grid over it. At each point on this grid, I calculate the dominant angle of change, essentially how the visual data flows or bends at that moment. These angle vectors give me a map of the image's inherent motion.
I then redraw the image by following these angles, line by line. It's not tracing. It's interpreting. The result is a set of flowing strokes that, in places, echo the original image's shapes and contours.
To further distill the image, I apply a color quantization algorithm that reduces the palette to just 4 to 8 tones. What was once a sea of millions of colors becomes a landscape of bold contrasts. This constraint forces simplification. Boundaries sharpen. Shapes emerge.
Sometimes, I also sample colors directly from the image to retain a more realistic or grounded palette. This creates a dynamic tension between abstraction and representation.
Flat regions, where the algorithm finds little directional change, become still, undisturbed planes of color. Active zones, where flow is detected, fill with directional line-work. The result is a depth effect reminiscent of relief printmaking techniques like woodblock or linocut, where raised surfaces receive ink and carve out negative space.
This combination of structured geometry and organic movement creates a tension that I find beautiful. By recreating the image with such limited information, just a few colors and directional flow, it forces a kind of perceptual reassembly. You're not seeing a flower or a leaf or a tree as it is. You're seeing how it moves, how it holds together, how it pushes and pulls against itself.
What I love most about this process is how it encourages reinterpretation. I've always been fascinated with the way we perceive the world, how each of us filters what we see through personal context, bias, memory. With this series, I wanted to explore that idea artistically. How can technology, especially AI, help us reconsider what's right in front of us?
I'm always amazed by how often I'll see something beautiful or strange in nature, a pattern in a leaf, the curve of a branch, the way light hits a rock, and when I point it out, someone else doesn't quite see what I see. That disconnect is fascinating to me. It reminds me how subjective perception truly is.
By starting with the familiar and breaking it apart, this series asks us to look again. It's about reframing the mundane and elevating it. About using machines not to replace creativity, but to extend it into new textures, new forms, new perspectives.
Through this project, I've aimed to capture more than just images—I've tried to create a system for rediscovery. One that mirrors how we often process the world: not as it is, but as it feels to move through it. As much as this is a technical exploration, it's also a personal one.
We all perceive the world through our own lens. This work is mine.