Research map / 3D vision / 3D generation
JanusMesh: Fast and Zero-Shot 3D Visual Illusion Generation via Cross-Space Denoising
Siang Ling Zhang, Huai-Hsun Cheng, Tsung-Ju Yang, 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 |
|---|---|---|---|
| use SDF-averaging cross-space denoising and view-conditioned texturing instead of direct concatenation | JanusMesh · 3D illusion prompt pairs | multi-object rate (%): 56 -> 18 | Improved |
| use SDF-averaging cross-space denoising instead of direct concatenation | JanusMesh · 3D illusion prompt pairs | boundary seam score: 1503.42 -> 798.31 | Improved |
| use JanusMesh instead of direct concatenation | JanusMesh · 3D illusion prompt pairs | CLIP similarity: 29.030 -> 28.170 | Hurt |
| use JanusMesh instead of direct concatenation | JanusMesh · 3D illusion prompt pairs | GPT accuracy (%): 76 -> 84 | Improved |
| use JanusMesh instead of SDS-based Shape From Semantic | JanusMesh · 3D illusion prompt pairs | runtime: ~40 min -> ~3-5 min | Improved |
| use JanusMesh instead of TRELLIS prompted with front/back text | JanusMesh · 3D illusion prompt pairs | FID: 174.130 -> 185.555 | Hurt |
| replace SDF averaging with direct latent averaging (ablation) | JanusMesh stage 1 · 3D illusion prompt pairs | geometry quality | Hurt |
| add view-conditioned texture synthesis (stage 2) (ablation) | JanusMesh · 3D illusion prompt pairs | semantic coherence at target views | Improved |
| add CLIP-guided orientation search instead of fixed 0/180 angles (ablation) | JanusMesh · 3D illusion prompt pairs | semantic recognizability / user preference: 91% preferred | Mixed |
| add noise guidance (blending or space control) (ablation) | JanusMesh · 3D illusion prompt pairs | geometric fusion quality | Mixed |
| use SDF average instead of union, blur average, Minkowski or polar blending (ablation) | JanusMesh stage 1 · 3D illusion prompt pairs | geometric integrity | Improved |
Closest papers by meaning
- ViewDiff: 3D-Consistent Image Generation with Text-to-Image Models (CVPR 2024 · similarity 0.87)
- Ouroboros3D: Image-to-3D Generation via 3D-aware Recursive Diffusion (CVPR 2025 · similarity 0.87)
- Cycle3D: High-quality and Consistent Image-to-3D Generation via Generation-Reconstruction Cycle (AAAI 2025 · similarity 0.87)
- Unique3D: High-Quality and Efficient 3D Mesh Generation from a Single Image (NeurIPS 2024 · similarity 0.87)
- One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion (CVPR 2024 · similarity 0.87)
- HIFA: High-fidelity Text-to-3D Generation with Advanced Diffusion Guidance (ICLR 2024 · similarity 0.87)
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.
