Research map / 3D vision / 3D generation
VistaDream: Sampling multiview consistent images for single-view scene reconstruction
Haiping Wang, Yuan Liu, Ziwei Liu, Wenping Wang, Zhen Dong, Bisheng Yang
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 Multiview Consistency Sampling refinement (ablation) | VistaDream coarse · 11 RealmDreamer images | LLaVA-IQA Edge: 0.285 -> 0.342 | Improved |
| add Multiview Consistency Sampling refinement (ablation) | VistaDream coarse · 11 RealmDreamer images | LLaVA-IQA Structure: 0.542 -> 0.611 | Improved |
| add Multiview Consistency Sampling refinement (ablation) | VistaDream coarse · 11 RealmDreamer images | LLaVA-IQA Noise-Free: 0.909 -> 0.951 | Improved |
| add Multiview Consistency Sampling refinement (ablation) | VistaDream coarse · 11 RealmDreamer images | LLaVA-IQA Quality: 0.709 -> 0.733 | Improved |
| add Multiview Consistency Sampling refinement (ablation) | VistaDream coarse · 11 RealmDreamer images | LLaVA-IQA Detail: 0.967 -> 0.951 | Hurt |
| use VistaDream instead of training-based CAT3D | CAT3D · 11 RealmDreamer images | LLaVA-IQA Structure: 0.464 -> 0.611 | Improved |
| use VistaDream instead of training-based CAT3D | CAT3D · 11 RealmDreamer images | LLaVA-IQA Quality: 0.765 -> 0.733 | Hurt |
| add global 3D scaffold from zoom-out inpainting (ablation) | VistaDream · single-view scenes | qualitative | Improved |
| use LLaVA descriptions instead of BLIP2 captions for scaffold inpainting (ablation) | VistaDream · single-view scenes | qualitative | Improved |
| use MCS instead of SDS for refinement (ablation) | VistaDream · single-view scenes | qualitative | Improved |
| raise consistency weight w_t from 0 to between 0.3 and 0.8 (ablation) | MCS · single-view scenes | qualitative | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- CAT3D — CAT3D: Create Anything in 3D with Multi-View Diffusion Models (NeurIPS 2024) · better on 2 of 5 shared results
- GenWarp — GenWarp: Single Image to Novel Views with Semantic-Preserving Generative Warping (NeurIPS 2024) · better on 5 of 5 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
- SyncDreamer: Generating Multiview-consistent Images from a Single-view Image (ICLR 2024 · similarity 0.89)
- MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024 · similarity 0.89)
- Sparse3D: Distilling Multiview-Consistent Diffusion for Object Reconstruction from Sparse Views (AAAI 2024 · similarity 0.89)
- Efficient-3Dim: Learning a Generalizable Single-image Novel-view Synthesizer in One Day (ICLR 2024 · similarity 0.88)
- Denoising Diffusion via Image-Based Rendering (ICLR 2024 · similarity 0.88)
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 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.
