Zyme is a recursive image ecology — a closed-loop system that continuously generates, resamples, and transforms images through iterative feedback.
Every 75 seconds, the system selects source images from its own output history, blends them through a diffusion pipeline with randomized parameters, and feeds the result back into the pool. The process runs continuously, accumulating an evolving visual archive.
The image on the left is the most recent output. The terminal on the right shows the parameters and lineage of the current generation cycle. Everything updates automatically.
This is not generative AI in the conventional sense — there are no text prompts describing a desired outcome. The system is self-referential, drawing only from its own history to produce emergent visual forms.