Surflo: Consistent 3D Surface Flow Model with Global State

2026-06-11

AI Summary

Existing feedforward 3D reconstruction methods struggle to simultaneously handle multi-view geometric redundancy, support arbitrary output resolutions, and ensure surface consistency. This work proposes a novel framework that encodes an arbitrary number of pose-free RGB images into a global implicit latent state and reconstructs 3D surfaces at any resolution by independently transporting noise points to the surface via flow matching. It is the first feedforward approach to integrate a global implicit representation with resolution-agnostic decoding, further enhancing local geometric consistency through photometric gradient guidance during ODE integration. Experiments demonstrate that the method matches or surpasses state-of-the-art feedforward techniques in surface reconstruction quality while achieving an order-of-magnitude faster inference compared to optimization-based approaches requiring hundreds of views.

Abstract

Geometry is invariant to viewpoint, which makes any collection of images a redundant encoding of a single 3D state. Existing feed-forward reconstruction models fail to exploit this: per-view methods emit overlapping, unaligned pointmaps that grow linearly with input count, while global-latent methods commit to a fixed, low-resolution output. We introduce Surflo, which compresses a variable number of unposed RGB views into K latent tokens-one global state-and decodes oriented 3D surface points by independently transporting them from noise onto the surface via flow matching. This frees the output from any fixed grid or token budget: the same latent yields from a few thousand to a million points in a single forward pass. To suppress the local inconsistencies inherent to independent per-point decoding, an inference-time guidance term correlates nearby points by injecting a photometric gradient during ODE integration. Surflo matches or surpasses feed-forward baselines on surface metrics, runs an order of magnitude faster than optimization-based methods that require hundreds of views, and is the only feed-forward approach to combine a global latent with arbitrary-resolution decoding.

Problem

• 3D surface reconstruction
• viewpoint invariance
• global latent representation
• arbitrary-resolution decoding
• geometric consistency

Innovation

• flow matching
• global latent representation
• arbitrary-resolution decoding
• 3D surface reconstruction
• photometric guidance