Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces HyperbolicDiffusion, a novel method for generating visual fields on the hyperbolic plane without the need for training. By employing a compact dynamic programming approach called Hyperbolic Blooming Cover, the method efficiently manages window placements and integrates local predictions from a standard diffusion model onto a shared latent canvas. The approach effectively addresses the challenges posed by curvature-induced discrepancies at multi-window junctions, resulting in sharp and consistent visual outputs that can be reprojected across different viewpoints.
Generating sharp and consistent visuals on the hyperbolic plane is now achievable in seconds, challenging traditional diffusion methods constrained by rectangular canvases.
Planar tiled diffusion denoises overlapping windows of one rectangular canvas. The hyperbolic plane has no such canvas, and its area grows exponentially with radius. We introduce HyperbolicDiffusion, a training-free method for generating finite visual fields directly on the hyperbolic plane H2. Our Hyperbolic Blooming Cover reduces window placement to a compact dynamic program that runs in seconds while providing strong theoretical guarantees. Permanent surface IDs form a shared latent canvas: a standard diffusion model denoises local windows, whose predictions are fused back onto H2. Because curvature causes residual disagreement and blur at multi-window junctions, a geometry-derived second stage re-noises and repairs precisely those regions. The resulting fields are sharp, reprojectable, and consistent across viewpoints, providing a prompt-driven generative counterpart to Escher's Circle Limit series.