Synthetic photography explorations in digital latent space.

The artwork were created using a custom, diffusion-based generative model, which transforms random noise into a detailed image through a learned denoising process. The diffusion model used in creating the images is fine-tuned using my own photos taken in 2008 - 2024.

The model is then manipulated through custom settings, glitching and blending algorithmic precision with creative unpredictability. The output images are formed within the digital latent space of the diffusion model architecture, resembling the process of latent imagery in analog photography. The final images resemble photos but are synthetic, digitally produced.

Generative animation made by creative coding in p5.js.

The base idea of the artwork revolves around topographic maps that are generated by experimental generative particle tracing. Generation of the artwork is randomized, so that every iteration of the animation is unique.

This series explores using generative artificial intelligence along with procedural image generation.  

The initial, mycelium-like pattern is generated using creative coding and then rendered using a custom diffusion AI model.
The final images are then hyper-upscaled multiple times to 12k resolution with small details generated in every upscale iteration.

Like the human mind, artificial intelligence also represents a world that exists 'out there,' beyond human brain activity or the software created to imitate it.
As users of AI and as artists, we can influence this representation by breaking, modifying, and questioning it.

What kind of reality do we want to represent through machine learning? How could machine learning be used to discover and visualize new things that are still invisible to us? Could generative AI be used to imagine or dream of speculative, optimistic, impossible, or completely unfamiliar realities that lie beyond the scope of our current visions of the future? Or could new technologies help us find something new within the reality that already exists?

Unimuotokuvia is a participatory artwork created in collaboration with municipality of Tuusula, Finland. The artwork is a 15 minute frame-by-frame animation generated using custom-trained diffusion AI models based on photographs of volunteers that took part in creating the artwork.

Showreel showcasing a selection of various 3D visuals including procedural simulation, animation and environment design.

Software used: SideFX Houdini, Substance Painter, Blender and Unreal Engine.

Digital still life explorations created using various 3D applications.