Home▸Generative Art & Algorithmic Patterns▸Ai Art Simulation

🧪 Ai Art Simulation

Interactive generative-art simulator: walk a 2D latent vector through a hand-built decoder function and watch a fractal-noise 'artwork' morph continuously, with a 3D latent-manifold view showing exactly where you are and a flow-field brush overlay driven by the same field.

Generative Art & Algorithmic Patterns2DModerate60 FPS📱 Mobile-adapted

Generative art tools — GANs, VAEs, diffusion decoders — all work by mapping a compact latent vector to a full image, and moving smoothly through that latent space produces smoothly morphing artwork. This simulator makes that idea tangible with a fully inspectable, hand-built decoder: a domain-warped fractal-noise field standing in for a trained network. Drag the z₁/z₂ sliders (or press Animate Latent Walk) to move a marker across a live 3D latent manifold while a companion canvas continuously redecodes the current vector into abstract art, and watch a flow-field "brush" of particles trace the decoder's own gradient structure in real time. Style-warp and detail sliders act as extra semantic directions, and Reseed Decoder swaps in a different noise seed to show how a differently-trained network changes the whole manifold.

⚙ Under the hood

This simulation demonstrates the use of artificial intelligence in generating artwork. It explores how algorithms can be trained to produce novel images based on various inputs and parameters, illustrating concepts related to generative art and machine learning.

AI ArtGenerative Design

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install