Research map / 3D vision / 3D generation
Cascade-Zero123: One Image to Highly Consistent 3D with Self-Prompted Nearby Views
Yabo Chen, Jiemin Fang, Yuyang Huang, Taoran Yi, Xiaopeng Zhang, Lingxi Xie, Xinggang Wang, Wenrui Dai, Hongkai Xiong, Qi 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 |
|---|---|---|---|
| use self-prompted nearby views with a cascaded Base and Refiner Zero-1-to-3 | Zero123-XL · Objaverse test split | PSNR: 18.68 -> 21.42 | Improved |
| apply the cascade to Zero123-XL on RTMV complex scenes | Zero123-XL · RTMV | PSNR: 10.97 -> 11.12 | Improved |
| apply the cascade to Zero123-XL on RealFusion15 | Zero123-XL · RealFusion15 | PSNR: 25.220 -> 26.098 | Improved |
| simply cascade two untrained Zero-1-to-3 models with self-prompted views (ablation) | Zero123-XL · Objaverse test (ablation) | PSNR: 18.68 -> 17.67 | Hurt |
| use nearby prompt views instead of random or larger-angle views (ablation) | Refiner-0123 · Objaverse test (ablation) | PSNR | Improved |
| include coarse target view among self-prompted views (ablation) | Refiner-0123 · Objaverse test (ablation) | PSNR | Improved |
| update Base-0123 with EMA from Refiner-0123 instead of keeping it fixed (ablation) | Cascade-Zero123 · Objaverse test (ablation) | PSNR | Improved |
| feed only a single image to the trained Refiner-0123 (ablation) | Refiner-0123 · Objaverse test (ablation) | PSNR: 18.68 -> 19.39 | Improved |
| apply the cascade to SyncDreamer | SyncDreamer · GSO novel view synthesis | PSNR/SSIM/LPIPS/CLIP | Improved |
| use cascade generation of views | Zero-1-to-3 inference cost · 100-view generation | inference time (s): 2203 -> 2356 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 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 2 of 2 shared results
- SyncDreamer — SyncDreamer: Generating Multiview-consistent Images from a Single-view Image (ICLR 2024) · better on 2 of 2 shared results
- Zero-1-to-3 — Zero-1-to-3: Zero-shot One Image to 3D Object (ICCV 2023) · better on 0 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
- ZeroNVS: Zero-Shot 360-Degree View Synthesis from a Single Image (CVPR 2024 · similarity 0.87)
- ZeroShape: Regression-based Zero-shot Shape Reconstruction (CVPR 2024 · similarity 0.87)
- MVGenMaster: Scaling Multi-View Generation from Any Image via 3D Priors Enhanced Diffusion Model (CVPR 2025 · similarity 0.86)
- MVD-Fusion: Single-view 3D via Depth-consistent Multi-view Generation (CVPR 2024 · similarity 0.86)
- Zero-to-Hero: Enhancing Zero-Shot Novel View Synthesis via Attention Map Filtering (NeurIPS 2024 · similarity 0.86)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · 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.
