Reproduce “Domain-Tuned Diffusion Super-Resolution for 4K Facial Asset Maps” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Elly Akhoundi, Mathieu Lamarre, Thibault Goyallon, Patrick Anderson, Étienne Danvoye

The paper refines 1K facial albedo and displacement maps to 4K. It bilinearly enlarges each map, then adapts a frozen diffusion model with LoRA and Tile ControlNet conditioning, separately for each modality. On 100 held-out maps, albedo LPIPS falls from 0.21 to 0.08; displacement detail energy reaches 97%, while alignment matches bilinear and pores are not recovered exactly.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.