SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation
SILSA generates high-resolution 3D shapes from compact slice latents while preserving topology and cutting token cost.
SILSA represents 3D shapes with compact overlapping slice latents along three axes instead of expensive voxel tokens, enabling single-stage rectified-flow generation. A Slice VAE and Volumetric Anchor Lattice preserve cross-sectional continuity, while persistence-diagram and Betti supervision enforce topology. Versus the strongest baseline, SILSA improves PSNR by 8.7%, coverage by 5.96 points, and Betti error by 9.2%, while using at least 70% fewer tokens and cutting inference time by 58.5%.
- Overlapping slice latents replace dense voxel tokens
- Topology supervision matches persistence diagrams and Betti transitions
- PSNR rises 8.7% and inference time falls 58.5%
- Token count is 70% below the next-most compact baseline
From the original article
1 min read at huggingface.co
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.
By Tianjiao Yu, Xinzhuo Li, Yifan Shen · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.