Serendipitous Swarm 2024.

The past few years have been a rabbit hole of epic proportions as far as creativity goes. I have been exploring machine learning on my own catalogue of art for quite some time now and despite the general lack of updates, I have been busy behind the scenes these last few years.

One approach to using Stable Diffusion in my artistic practice has been making models of my art. Some models are based on single pieces and others are combinations of multiple and sometimes many artworks. Using these models I am then able to produce new images in my style, often traced directly over old pieces or new un-rendered Blender creations.

Using batch combination I can then recombine the resulting image sets into further image sets and then repeat the process all over again. I’ve currently done 7 generations of this process and I’m starting to get the kinds of results I went into this to find.

Check out this gallery which has a lot of curated and some random images in it: https://www.beaudeeley.net/

Grab some LORAs I’ve made if you’re on CivitAI: https://civitai.com/user/Xenodimensional


2026 Update:

While I feel that what I was pursuing was ethical as I was using my own artworks as inputs and local generation, I have essentially pivoted away from using machine learning and AI in my artistic practice and instead been focussing on airbrushing, laser cutting and electronics. Digital art was always a hard sell and since the rise of AI image generation tools it has become impossible to justify the time spent on digital art when it is instantly lost in the ocean of images.

Input image example, made in Blender

2023 example

2024 example