About
Contact
CV
Artificial Intelligence and Generative Art Experiments
Experimental, artificial intelligence, generative art.
When Dall·E came out I became obsessed with the idea that I could get to the eureka moment of seeing a visual idea come to life in an instant and without the effort it previously took.
As a designer, I’ve seen AI erode my bottom line early on and I knew I needed to at least figure out what’s going on.
Below are some experiments of me trying to figure out exactly
— what’s going on.
Nnnnound
Google AI Studio, Gemini, Claude, Nano Banana, Ideogram
First attempt at vibe coding. The name is a combination of ffffound.com (a now-defunct cult inspiration tool of the early-to-mid 2010s) and Nano Banana (Google’s image generation model). I wanted to see if I could recreate the feeling of ffffound.com with images that don’t yet exist.
Can an algorithm show you what you want to see before you even thought it? Before it even exists?
I am trying to develop a complex genome-based algorithim that tracks a user’s interests and builds a taste profile as the user dives deeper into a rabbit hole. The system feeds the signals back into itself and builds prompts from everything it knows, without the user having to write a single prompt.
Still an open-ended question. I have moved the coding over to Claude Code, and the generations to Ideogram 4.0.
Preview available upon request.
Nnnnound Preview
Anthopic Claude, PyTorch
A mini-site that showcases over 1000 generations made using the Nnnnound prototype.
All the images went through an image encoder to precompute their eight nearest neighbours. This allowes me to show them as “suggested images.”
The frontend is an infinite-scroll masonry grid, and a lightweight client-side “affinity” model keeps track of the user’s clicks to rank their preferences and quietly personalize the browsing experience with no server calls.
Visit the Live Preview
LexiGenome
Google AI Studio, Anthropic Claude
A friend of mine used to love the word “serendipity,” which in turn made me aware of how special some words are. Later on I discovered wabi-sabi, hygge, and schadenfreude – now all part of my top 10 favourite words. As I was beginning to think about algorithms, I thought this could be a fun way for me to get a feel for genome-based algorithms in a simple way and to add to my top words list.
Gemini and Google AI Studio helped me one-shot a prototype and iterate on its functionality. The tool guides the user along a journey of linguistic discovery, tying words together over 7 axis, and a secret “Surprise Me” 8th.
Preview available upon request.
Ideogram
Anthropic Claude, Ideogram
I am currently the Creative Curation Lead at Ideogram, which means a lot of things to a lot of people. I’m the visual person and I use my design and visual arts experience to advise on the inputs and outputs of the state of the art models the company produces.
36 Days of Type
Stable Diffusion, Controlnet
36 Days of Type was a yearly social design challenge involving creating a letter form each day for 36 days going through the 26 letters of the alphabet and the 10 digits. I wanted to try using generative art for it but early experiments proved that AI didn’t understand letters (at least not at the time – today there are plenty of generative tools able to type set complete sentences).
I found instead a way to generate images based on grayscale z-depth maps, as a sort of reverse logic to that of portrait photos on the iPhone.
I designed the letters in grayscale, using the lightness of the pixels to indicate the “distance” from the camera, as a reverse depth map. I was able to feed these back to the Stable Diffusion model as a starting point.
From there, I was able to prompt just about anything and the desired shape prevailed through each iteration.
For reference, swipe to see all the depth maps I created. My idea was to design midcentury modern lamps, and I used that as a part of my prompts in most cases in the above examples.
An added bonus of the depth maps was the ability to use it as z-depth information for video as well. Music was generated using Google’s MusicLM was the cherry on top.