Research map / 3D vision / 3D generation
3D Photo Stylization: Learning To Generate Stylized Novel Views From a Single Image
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| jointly model stylization and view synthesis using a shared featurized point cloud | Ours versus sequential 3DPhoto + style-transfer baselines · 3D photo stylization consistency | RMSE: 0.092 -> 0.086 | Helped |
| jointly model stylization and view synthesis using a shared featurized point cloud | Ours versus sequential 3DPhoto + style-transfer baselines · 3D photo stylization consistency | LPIPS: 0.200 -> 0.133 | Helped |
| add a multi-view consistency loss during view synthesis and stylization training (ablation) | 3D photo stylization model · stylized novel views | LPIPS: 0.199 -> 0.128 | Helped |
| add a multi-view consistency loss during view synthesis and stylization training (ablation) | 3D photo stylization model · stylized novel views | RMSE: 0.113 -> 0.086 | Mixed |
| add a consistency loss in the view synthesis stage only (ablation) | 3D photo stylization model · stylized novel views | RMSE: 0.113 -> 0.109 | Helped |
| add a consistency loss in the view synthesis stage only (ablation) | 3D photo stylization model · stylized novel views | LPIPS: 0.199 -> 0.190 | Helped |
| add a consistency loss in the stylization stage only (ablation) | 3D photo stylization model · stylized novel views | RMSE: 0.113 -> 0.081 | Helped |
| add a consistency loss in the stylization stage only (ablation) | 3D photo stylization model · stylized novel views | LPIPS: 0.199 -> 0.132 | Helped |
| learn content features directly on the point cloud instead of projecting VGG image features (ablation) | 3D photo stylization model · stylized novel views | visual artifacts | Helped |
| normalize the point cloud to NDC before feature processing (ablation) | 3D photo stylization model · stylized novel views | stylization quality | Helped |
| use the proposed local-input point-cloud approach rather than StyleScene global-input | multi-view stylization · Tanks and Temples, Truck, short-range | RMSE: 0.124 -> 0.099 | Helped |
| use the proposed local-input point-cloud approach rather than StyleScene global-input | multi-view stylization · Tanks and Temples, Truck, long-range | LPIPS: 0.188 -> 0.128 | Helped |
Closest papers by meaning
- Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles (NeurIPS 2025 · similarity 0.91)
- 3DToonify: Creating Your High-Fidelity 3D Stylized Avatar Easily from 2D Portrait Images (CVPR 2024 · similarity 0.90)
- FPRF: Feed-Forward Photorealistic Style Transfer of Large-Scale 3D Neural Radiance Fields (AAAI 2024 · similarity 0.88)
- PNeSM: Arbitrary 3D Scene Stylization via Prompt-Based Neural Style Mapping (AAAI 2024 · similarity 0.88)
- Morpheus: Text-Driven 3D Gaussian Splat Shape and Color Stylization (CVPR 2025 · similarity 0.88)
- ReGS: Reference-based Controllable Scene Stylization with Gaussian Splatting (NeurIPS 2024 · similarity 0.87)
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.