Research map / 3D vision / Face and shape
Voxify3D: Pixel Art Meets Volumetric Rendering
Yi-Chuan Huang, Jiewen Chan, Hao-Jen Chien, Yu-Lun Liu
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 stage 1 coarse voxel initialization (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 28.42 -> 40.06 | Improved |
| add stage 2 orthographic pixel art fine-tuning (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 34.32 -> 40.06 | Improved |
| use orthographic instead of perspective projection (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 27.38 -> 40.06 | Improved |
| add depth loss (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 39.75 -> 40.06 | Improved |
| add patch CLIP loss (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 39.23 -> 40.06 | Improved |
| add Gumbel-Softmax palette quantization (ablation) | Voxify3D · CLIP-IQA ablation (5 objects) | CLIP-IQA: 39.31 -> 40.06 | Improved |
| add Gumbel-Softmax quantization (ablation) | Voxify3D · Expert color preference (10 art-trained) | preferred (%): 11.11 -> 88.89 | Improved |
| Voxify3D vs strongest baseline Blender | mesh-to-voxel-art methods · CLIP-IQA (35 examples) | CLIP-IQA: 36.31 -> 37.12 | Improved |
| increase palette size from 2 to 8 colors (ablation) | Voxify3D · qualitative palette study | color richness / detail | Improved |
Closest papers by meaning
- QUADify: Extracting Meshes with Pixel-level Details and Materials from Images (CVPR 2024 · similarity 0.88)
- MeshGen: Generating PBR Textured Mesh with Render-Enhanced Auto-Encoder and Generative Data Augmentation (CVPR 2025 · similarity 0.87)
- MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers (ICLR 2025 · similarity 0.87)
- DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior (ICLR 2024 · similarity 0.87)
- LiteReality: Graphic-Ready 3D Scene Reconstruction from RGB-D Scans (NeurIPS 2025 · similarity 0.86)
- GeoSVR: Taming Sparse Voxels for Geometrically Accurate Surface Reconstruction (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.
