Native and Compact Structured Latents for 3D Generation

Papers with Code Trending 12/16/25, 06:58 PM Papers

3d-generation sparse-voxel generative-modeling flow-matching computer-vision vae

Summary

This paper introduces O-Voxel, a new sparse voxel representation for 3D generative modeling that efficiently handles complex topologies and appearance, and trains large-scale flow-matching models with 4B parameters to achieve state-of-the-art generation quality.

Recent advancements in 3D generative modeling have significantly improved the generation realism, yet the field is still hampered by existing representations, which struggle to capture assets with complex topologies and detailed appearance. This paper present an approach for learning a structured latent representation from native 3D data to address this challenge. At its core is a new sparse voxel structure called O-Voxel, an omni-voxel representation that encodes both geometry and appearance. O-Voxel can robustly model arbitrary topology, including open, non-manifold, and fully-enclosed surfaces, while capturing comprehensive surface attributes beyond texture color, such as physically-based rendering parameters. Based on O-Voxel, we design a Sparse Compression VAE which provides a high spatial compression rate and a compact latent space. We train large-scale flow-matching models comprising 4B parameters for 3D generation using diverse public 3D asset datasets. Despite their scale, inference remains highly efficient. Meanwhile, the geometry and material quality of our generated assets far exceed those of existing models. We believe our approach offers a significant advancement in 3D generative modeling.

Original Article

View Cached Full Text

Cached at: 07/25/26, 05:03 AM

Paper page - Native and Compact Structured Latents for 3D Generation

Source: https://huggingface.co/papers/2512.14692 Published on Dec 16, 2025

Authors:

Abstract

A new sparse voxel representation called O-Voxel enables high-quality 3D generative modeling with efficient inference and robust topology handling.

Recent advancements in 3D generative modeling have significantly improved the generation realism, yet the field is still hampered by existing representations, which struggle to capture assets with complex topologies and detailedappearance. This paper present an approach for learning a structured latent representation from native 3D data to address this challenge. At its core is a newsparse voxel structurecalledO-Voxel, anomni-voxel representationthat encodes bothgeometryandappearance.O-Voxelcan robustly model arbitrary topology, including open, non-manifold, and fully-enclosed surfaces, while capturing comprehensive surface attributes beyond texture color, such asphysically-based rendering parameters. Based onO-Voxel, we design aSparse Compression VAEwhich provides a high spatial compression rate and a compact latent space. We train large-scaleflow-matching modelscomprising 4B parameters for 3D generation using diverse public 3D asset datasets. Despite their scale, inference remains highly efficient. Meanwhile, thegeometryand material quality of our generated assets far exceed those of existing models. We believe our approach offers a significant advancement in 3D generative modeling.