Research map / 3D vision / 3D generation
Make-It-3D: High-fidelity 3D Creation from A Single Image with Diffusion Prior
Junshu Tang, Tengfei Wang, Bo Zhang, Ting Zhang, Ran Yi, Lizhuang Ma, Dong Chen
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 texture-enhancement refine stage (textured point cloud) after coarse NeRF (ablation) | Make-It-3D · Test benchmark (400 images) | LPIPS: 0.2354 -> 0.0780 | Improved |
| add refine stage on DTU (ablation) | Make-It-3D · DTU | CLIP score: 87.50% -> 95.65% | Improved |
| use timestep-split SDS/CLIP-D (threshold 400) instead of simultaneous SDS+CLIP-D (ablation) | Make-It-3D coarse stage · Test benchmark (400 images) | Contextual distance: 2.32 -> 1.74 | Improved |
| optimize SDS and CLIP-D simultaneously rather than switching by timestep (ablation) | Make-It-3D coarse stage · Test benchmark (400 images) | CLIP score: 87.50% -> 84.01% | Hurt |
| use only SDS loss without CLIP-D (ablation) | Make-It-3D coarse stage · Test benchmark (400 images) | LPIPS: 0.1427 -> 0.3045 | Hurt |
| use only CLIP-D loss without SDS (ablation) | Make-It-3D coarse stage · Test benchmark (400 images) | CLIP score: 87.50% -> 80.27% | Hurt |
| vary timestep threshold between SDS and CLIP-D (300, 400, 500) (ablation) | Make-It-3D coarse stage · Test benchmark (400 images) | Contextual distance: 2.19 / 1.74 / 2.23 | Mixed |
| add depth prior (Pearson correlation) at reference view (ablation) | Make-It-3D coarse stage · qualitative | geometry plausibility | Improved |
| use learnable point descriptors and NeRF texture initialization in refine stage (ablation) | Make-It-3D refine stage · qualitative | texture quality | Improved |
| sample timesteps only from 200 to 600 in SDS (ablation) | Make-It-3D coarse stage · qualitative | geometry and alignment | Improved |
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.
- SinNeRF — SinNeRF: Training Neural Radiance Fields on Complex Scenes from a Single Image (ECCV 2022) · 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
- Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors (ICLR 2024 · similarity 0.92)
- DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior (ICLR 2024 · similarity 0.91)
- Sculpt3D: Multi-View Consistent Text-to-3D Generation with Sparse 3D Prior (CVPR 2024 · similarity 0.90)
- One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion (CVPR 2024 · similarity 0.90)
- Neural Point Cloud Diffusion for Disentangled 3D Shape and Appearance Generation (CVPR 2024 · similarity 0.90)
- DIRECT-3D: Learning Direct Text-to-3D Generation on Massive Noisy 3D Data (CVPR 2024 · similarity 0.90)
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.
