SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation

Hugging Face Daily Papers Papers
3d-generation voxel-latents topology rectified-flow vae diffusion efficiency

Summary

SILSA is a topology-aware 3D generation framework that replaces expensive voxel tokens with compact sliding-window slice latents and single-stage rectified-flow generation, using persistence-diagram/Betti-based slice-level topology supervision. It improves structural fidelity (PSNR +8.7%, Betti error -9.2%) while cutting tokens by 70%, memory by 40.4%, and inference time by 58.5%.

High-resolution 3D generation increasingly relies on voxel latents and multi-stage pipelines that first predict active structure and then synthesize local geometry. While effective, this design fragments continuous surfaces into many local tokens, inflates generation cost, and often weakens topological consistency for thin or highly connected shapes. We introduce SILSA, a topology-aware 3D generation framework that represents shapes with compact sliding-window slice latents. Instead of generating expensive voxel tokens, SILSA uses a fixed set of overlapping slices along the three canonical axes, where each token summarizes a local depth window to preserve cross-sectional continuity and support single-stage rectified-flow generation. A Slice VAE encodes oriented surface samples into multi-axis slice latents and reconstructs them with a sparse volumetric decoder, while a Volumetric Anchor Lattice coordinates directional slice streams through a shared 3D workspace. To preserve structural correctness, we introduce slice-level topology supervision that matches persistence diagrams and aligns Betti transitions across neighboring slices. Experiments show that SILSA improves structural fidelity while substantially reducing generation cost. SILSA improves PSNR by 8.7%, coverage by 5.96 absolute points, and Betti error by 9.2% over the strongest baseline, while using 70.0% fewer tokens than the next-most compact baseline and over 98% fewer tokens than sparse or hierarchical tokenizers, effectively reducing training memory by 40.4% and inference time by 58.5%. Qualitative results further show improved preservation of thin structures, repeated components, and long-range connectivity.
Original Article
View Cached Full Text

Cached at: 10/02/26, 08:28 AM

Paper page - SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation

Source: https://huggingface.co/papers/2610.02201

Abstract

High-resolution3Dgenerationincreasinglyreliesonvoxellatentsandmulti-stagepipelinesthatfirstpredictactivestructureandthensynthesizelocalgeometry.Whileeffective,thisdesignfragmentscontinuoussurfacesintomanylocaltokens,inflatesgenerationcost,andoftenweakenstopologicalconsistencyforthinorhighlyconnectedshapes.WeintroduceSILSA,atopology-aware3Dgenerationframeworkthatrepresentsshapeswithcompactsliding-windowslicelatents.Insteadofgeneratingexpensivevoxeltokens,SILSAusesafixedsetofoverlappingslicesalongthethreecanonicalaxes,whereeachtokensummarizesalocaldepthwindowtopreservecross-sectionalcontinuityandsupportsingle-stagerectified-flowgeneration.ASliceVAEencodesorientedsurfacesamplesintomulti-axisslicelatentsandreconstructsthemwithasparsevolumetricdecoder,whileaVolumetricAnchorLatticecoordinatesdirectionalslicestreamsthroughashared3Dworkspace.Topreservestructuralcorrectness,weintroduceslice-leveltopologysupervisionthatmatchespersistencediagramsandalignsBettitransitionsacrossneighboringslices.ExperimentsshowthatSILSAimprovesstructuralfidelitywhilesubstantiallyreducinggenerationcost.SILSAimprovesPSNRby8.7%,coverageby5.96absolutepoints,andBettierrorby9.2%overthestrongestbaseline,whileusing70.0%fewertokensthanthenext-mostcompactbaselineandover98%fewertokensthansparseorhierarchicaltokenizers,effectivelyreducingtrainingmemoryby40.4%andinferencetimeby58.5%.Qualitativeresultsfurthershowimprovedpreservationofthinstructures,repeatedcomponents,andlong-rangeconnectivity.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2610\.02201

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/2610.02201 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.02201 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2610.02201 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

Structured 3D Latents for Scalable and Versatile 3D Generation

Papers with Code Trending

微软提出统一的结构化潜在表示（SLAT），结合 rectified flow transformers 实现高质量、多格式（Radiance Fields、3D Gaussians、网格）的 3D 资产生成，支持文本/图像条件及局部编辑，在 500K 3D 数据集上训练的 2B 参数模型显著超越现有方法。

Native and Compact Structured Latents for 3D Generation

Papers with Code Trending

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.

FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generation

Hugging Face Daily Papers

FLAT proposes a method to decode explicit triangle splats directly from video diffusion latents for geometrically accurate 3D scene generation. It introduces a ray-centered rotation parameterization and a product window function to improve gradient flow, achieving better geometric accuracy than prior feedforward methods while supporting real-time rendering.

TriSplat: Simulation-Ready Feed-Forward 3D Scene Reconstruction

Hugging Face Daily Papers

TriSplat is a feed-forward 3D reconstruction network that uses oriented triangle primitives to directly generate simulation-ready meshes from single images, bypassing expensive post-processing steps. It achieves geometry-faithful reconstructions while maintaining competitive novel-view rendering quality.