Research map / 3D vision / 3D generation
HiFi-123: Towards High-fidelity One Image to 3D Content Generation
Wangbo Yu, Li Yuan, Yan-Pei Cao, Xiangjun Gao, Xiaoyu Li, Wenbo Hu, Long Quan, Ying Shan, Yonghong Tian
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 |
|---|---|---|---|
| apply RGNV enhancement to Zero-1-to-3 novel views | Zero-1-to-3 · GSO novel view synthesis | PSNR: 18.95 -> 20.45 | Improved |
| apply RGNV enhancement to SyncDreamer novel views | SyncDreamer · single view dataset | CLIP similarity: 0.851 -> 0.896 | Improved |
| use RGSD loss instead of SDS loss in the refine stage (ablation) | HiFi-123 · GSO 3D dataset | PSNR: 19.25 -> 23.68 | Improved |
| use RGSD loss instead of SDS loss in the refine stage (ablation) | HiFi-123 · Single view dataset | Contextual distance: 1.925 -> 1.627 | Improved |
| refine stage with SDS loss over coarse stage (ablation) | HiFi-123 · Single view dataset | Contextual distance: 1.901 -> 1.925 | No clear effect |
| use depth-conditioned DDIM inversion rather than regular DDIM inversion (ablation) | RGNV · 400 images reconstruction | LPIPS: 0.2835 -> 0.0661 | Improved |
| remove attention injection from RGNV or use SDEdit instead (ablation) | RGNV · qualitative ablation | visual consistency | Hurt |
| use RGSD loss with alternating fixed-view targets for faster refinement | HiFi-123 refine stage · training time | minutes: 60 -> 10 | Improved |
| naive image loss using RGNV-enhanced rendered views (ablation) | HiFi-123 refine stage · qualitative | texture sharpness | Hurt |
| increase attention injection start step l (ablation) | RGNV · qualitative | texture fidelity vs geometry | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- Magic123 — Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors (ICLR 2024) · better on 5 of 5 shared results
- SyncDreamer — SyncDreamer: Generating Multiview-consistent Images from a Single-view Image (ICLR 2024) · better on 4 of 4 shared results
- Zero-1-to-3 — Zero-1-to-3: Zero-shot One Image to 3D Object (ICCV 2023) · better on 3 of 3 shared results
- RealFusion — RealFusion: 360deg Reconstruction of Any Object From a Single Image (CVPR 2023) · better on 2 of 2 shared results
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
- Cycle3D: High-quality and Consistent Image-to-3D Generation via Generation-Reconstruction Cycle (AAAI 2025 · similarity 0.89)
- One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion (CVPR 2024 · similarity 0.89)
- Diffusion$^2$: Dynamic 3D Content Generation via Score Composition of Video and Multi-view Diffusion Models (ICLR 2025 · similarity 0.88)
- ViewDiff: 3D-Consistent Image Generation with Text-to-Image Models (CVPR 2024 · similarity 0.88)
- IM-3D: Iterative Multiview Diffusion and Reconstruction for High-Quality 3D Generation (ICML 2024 · similarity 0.88)
- DreamTime: An Improved Optimization Strategy for Diffusion-Guided 3D Generation (ICLR 2024 · similarity 0.88)
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.
