Research map / 3D vision / Face and shape
Efficient Autoregressive Shape Generation via Octree-Based Adaptive Tokenization
Kangle Deng, Hsueh-Ti Derek Liu, Yiheng Zhu, Xiaoxia Sun, Chong Shang, Kiran S. Bhat, Deva Ramanan, Jun-Yan Zhu, Maneesh Agrawala, Tinghui Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add quadric-error adaptive subdivision to the octree tokenizer (ablation) | OAT (discrete) · Objaverse test | IoU: 84.2 -> 86.7 | Improved |
| add adaptive subdivision at similar token count (607 vs 625 tokens) (ablation) | OAT (discrete) · Objaverse test | IoU: 88.3 -> 89.7 | Improved |
| add adaptive subdivision at similar token count (1726 vs 1284 tokens) (ablation) | OAT (discrete) · Objaverse test | IoU: 89.9 -> 90.2 | Improved |
| add adaptive subdivision for continuous latents (607 vs 625 latents) (ablation) | OAT-KL · Objaverse test | IoU: 92.1 -> 92.7 | Improved |
| use OAT tokenization instead of fixed-size Craftsman-VQ at about half the tokens | OAT (discrete) · Objaverse test | IoU: 83.8 -> 88.6 | Improved |
| increase OAT token count from 266 to 1284 (ablation) | OAT (discrete) · Objaverse test | IoU: 86.7 -> 90.2 | Improved |
| replace Craftsman-VQ+GPT with OctreeGPT on OAT tokens | text-to-shape generation · Objaverse | FID: 85.10 -> 56.88 | Improved |
| use breadth-first instead of depth-first node ordering (ablation) | OctreeGPT · Objaverse | FID: 67.31 -> 56.88 | Improved |
| use breadth-first next-token instead of next scale prediction (ablation) | OctreeGPT · Objaverse | FID: 198.37 -> 56.88 | Improved |
| add hierarchical octree structure over leaf-only tokens (ablation) | OAT · Objaverse generation | KID: 13.65 -> 5.79 | Improved |
| use convolution pooling instead of average pooling in encoder (ablation) | OAT encoder · Objaverse | IoU: 88.5 -> 88.6 | No clear effect |
| use concat + MLP pooling instead of average pooling (ablation) | OAT encoder · Objaverse | IoU: 88.6 -> 88.6 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- XCube — XCube: Large-Scale 3D Generative Modeling using Sparse Voxel Hierarchies (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- LoST: Level of Semantics Tokenization for 3D Shapes (CVPR 2026 · similarity 0.88)
- SAR3D: Autoregressive 3D Object Generation and Understanding via Multi-scale 3D VQVAE (CVPR 2025 · similarity 0.86)
- Enhancing Implicit Shape Generators Using Topological Regularizations (ICML 2024 · similarity 0.86)
- Make-A-Shape: a Ten-Million-scale 3D Shape Model (ICML 2024 · similarity 0.86)
- ShapeLLM-Omni: A Native Multimodal LLM for 3D Generation and Understanding (NeurIPS 2025 · similarity 0.86)
- When Worse is Better: Navigating the Compression Generation Trade-off In Visual Tokenization (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
