Surflo: Consistent 3D Surface Flow Model with Global State

Hugging Face Daily Papers Papers

Summary

Surflo is a feed-forward 3D reconstruction model that compresses unposed RGB views into latent tokens and decodes consistent 3D surface points via flow matching, enabling variable-resolution output and outperforming existing methods in speed.

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.
Original Article
View Cached Full Text

Cached at: 06/12/26, 02:52 AM

Paper page - Surflo: Consistent 3D Surface Flow Model with Global State

Source: https://huggingface.co/papers/2606.13644

Abstract

Surflo compresses unposed RGB views into latent tokens and decodes 3D surface points through flow matching, enabling flexible resolution output and efficient processing compared to existing methods.

Geometry is invariant to viewpoint, which makes any collection of images a redundant encoding of a single 3D state. Existingfeed-forward reconstruction modelsfail to exploit this: per-view methods emit overlapping, unalignedpointmapsthat grow linearly with input count, whileglobal-latent methodscommit to a fixed, low-resolution output. We introduce Surflo, which compresses a variable number of unposed RGB views into Klatent tokens-one global state-and decodes oriented 3D surface points by independently transporting them from noise onto the surface viaflow 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 aphotometric gradientduringODE integration. Surflo matches or surpasses feed-forward baselines onsurface 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.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2606\.13644

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.13644 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.13644 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.13644 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SurGe: Improved Surface Geometry in Point Maps

Hugging Face Daily Papers

SurGe introduces a Neighborhood Attention Decoder and a reformulated scale-invariant gradient matching loss to improve local surface geometry accuracy in feedforward 3D reconstruction, particularly for thin structures. It achieves state-of-the-art average rank on zero-shot monocular geometry benchmarks, with better local point map and normal metrics.

Language Modeling with Hyperspherical Flows

arXiv cs.LG

This paper introduces S-FLM, a novel flow-based language model that operates in a hyperspherical latent space to address the computational costs and semantic limitations of existing discrete diffusion and continuous flow models.

Asymmetric Flow Models

Hugging Face Daily Papers

Asymmetric Flow Modeling (AsymFlow) restricts noise prediction to low-rank subspaces for efficient high-dimensional flow-based generation, achieving state-of-the-art results on ImageNet and text-to-image tasks by fine-tuning from latent flow models.