Research map / 3D vision / 3D generation
Infinite Nature: Perpetual View Generation of Natural Scenes From a Single Image
Andrew Liu, Richard Tucker, Varun Jampani, Ameesh Makadia, Noah Snavely, Angjoo Kanazawa
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 recurrent multi-step training (T=5 instead of T=1) (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | FID-50: 95.4 -> 50.6 | Improved |
| add recurrent multi-step training (T=5 instead of T=1) (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | LPIPS: 0.30 -> 0.32 | No clear effect |
| add recurrent multi-step training (T=5 instead of T=1) (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | MSE: 0.022 -> 0.020 | Improved |
| use SPADE refinement network instead of simple U-Net (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | FID-50: 183.0 -> 50.6 | Improved |
| use SPADE refinement network instead of simple U-Net (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | LPIPS: 0.54 -> 0.32 | Improved |
| use SPADE refinement network instead of simple U-Net (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | MSE: 0.052 -> 0.020 | Improved |
| add geometric grounding (re-grounding of disparity) (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | FID-50: 64.3 -> 50.6 | Improved |
| add geometric grounding (re-grounding of disparity) (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | LPIPS: 0.34 -> 0.32 | Improved |
| add geometric grounding for 500-frame generation (ablation) | Infinite Nature · ACID perpetual generation (500 frames) | FID-50 over time | Improved |
| add the refine step (ablation) | Infinite Nature · ACID short-to-medium range (279 sequences) | visual quality (qualitative) | Improved |
| apply SynSin/MPI iteratively at test time or retrain them with repeat | SynSin and single-view MPI · ACID short-to-medium range (279 sequences) | image quality | Hurt |
Closest papers by meaning
- InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Images (ECCV 2022 · similarity 0.91)
- Pseudo-Generalized Dynamic View Synthesis from a Video (ICLR 2024 · similarity 0.86)
- Novel View Synthesis from A Few Glimpses via Test-Time Natural Video Completion (NeurIPS 2025 · similarity 0.86)
- ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models (CVPR 2024 · similarity 0.85)
- World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.85)
- Generative Photography: Scene-Consistent Camera Control for Realistic Text-to-Image Synthesis (CVPR 2025 · similarity 0.85)
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.
