RaPD: Resolution-Agnostic Pixel Diffusion via Semantics-Enriched Implicit Representations

TL;DR

RaPD introduces a resolution-agnostic generative model that performs diffusion in a continuous neural image field, enabling high-quality, scale-aware image synthesis at arbitrary resolutions.

Contribution

It proposes a novel diffusion approach in a continuous neural field with semantic guidance and coordinate-based rendering, improving resolution scalability and generation quality.

Findings

Superior generation quality demonstrated across resolutions

Enables arbitrary resolution rendering with fixed diffusion cost

Bridges the gap between continuous rendering and discrete latent spaces

Abstract

Natural images are continuous, yet most generative models synthesize them on discrete grids, limiting resolution-flexible generation. Continuous neural fields enable resolution-free rendering, but prior methods introduce continuity only at the decoding stage as an interpolation module, leaving the generative latent space discretized and reconstruction-oriented. We propose RaPD (Resolution-agnostic Pixel Diffusion), which performs diffusion in a continuous Neural Image Field (NIF) latent space. RaPD bridges this reconstruction-generation gap with Semantic Representation Guidance for generation-aware latent learning and a Coordinate-Queried Attention Renderer for coordinate-conditioned, scale-aware rendering. A single denoised latent can be rendered at arbitrary resolutions by changing only the query coordinates, keeping diffusion cost fixed. Experiments demonstrate superior generation quality and resolution scalability.