Diffusion model collapse isn't all bad: new sampler extracts population atlases without retraining

kwangmoo_yi · x · 2026-09-29

A new arXiv paper (Shi, Femiani, Wonka), "Atlases Are Already Inside," introduces an inference-time sampler that gives pretrained diffusion models a capability they were never trained for: constructing the atlas of the population they synthesize.

Evaluated as a registration target, the intrinsic atlas is best or second-best on every dataset against classical and learned templates, and the most central template on held-out brain MRI cohorts. The paper reframes atlas construction as a byproduct of generative modeling.

Related event: Deliberately Collapsing Diffusion Models Recovers Population Atlases(2 posts)→

Original post →

More from Multimodal

Multimodal channel →