Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
SuperVoxelGPT addresses the limitations of existing 3D tokenization methods for autoregressive shape generation by introducing an adaptive supervoxel representation based on geometric saliency. It uses centroidal Voronoi tessellation to create fine-grained cells in complex regions and larger cells in smooth regions, thus reducing redundancy. Experiments on Trellis-500K demonstrate that SuperVoxelGPT achieves state-of-the-art generation quality with a 10x speedup and a significant reduction in token sequence length compared to uniform voxel tokenization.
Autoregressive 3D shape generation just got 10x faster and more accurate by ditching fixed voxel grids for adaptive supervoxels that focus on geometrically salient regions.
Autoregressive multimodal large language models (MLLMs) enable 3D generation but struggle to scale to high-resolution shapes due to inadequate 3D tokenizations. Compact set-based representations discard deterministic spatial ordering, leading to ambiguous sequence prediction, while uniform or octree-based voxel grids preserve ordering at the cost of severe redundancy and excessively long sequences. This structural trade-off limits stable and efficient autoregressive 3D generation. We present SuperVoxelGPT, a representation-first framework that resolves this tension through adaptive and deterministically ordered supervoxel tokenization. Given a prompt, we first predict a coarse geometric saliency distribution and construct a shape-adaptive supervoxel partition using saliency-guided centroidal Voronoi tessellation, allocating fine-grained cells to complex regions and larger cells to smooth regions. Conditioned on the text and ordered supervoxel layout, we introduce a SuperVoxelVAE and fine-tune a pretrained MLLM to autoregressively generate supervoxel tokens. Experiments on Trellis-500K show that SuperVoxelGPT reduces token sequence length to 12.8% of uniform voxel tokenization while achieving state-of-the-art generation quality and an average 10$\times$ speedup over prior methods.