Research map / 3D vision / Gaussian splatting
Capacity-Controlled Multi-View Stylization of 3D Gaussian Splatting
Zhihao Wen, Yixin Yang, Bojian Wu, Yang Zhou, Dani Lischinski, Danny Cohen-Or, Hui Huang
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 enhanced reconstruction (depth, shape, size regularizers) instead of original 3DGS (ablation) | SceneStyler · 3DGS stylization (60 test cases) | ArtFID: 24.133 -> 22.801 | Improved |
| add cross-view matching guidance (lambda_guide 0 -> 1) (ablation) | SceneStyler · Cross-view guidance ablation | Long MEt3R: 0.2807 -> 0.2795 | Improved |
| increase guidance weight from 1 to 10 (ablation) | SceneStyler · Cross-view guidance ablation | Structure Loss: 0.0318 -> 0.0323 | Hurt |
| add cross-view guidance (lambda_guide 0 -> 1) (ablation) | SceneStyler · Cross-view guidance ablation | Structure Loss: 0.0312 -> 0.0318 | Hurt |
| add cross-view guidance (lambda_guide 0 -> 1) (ablation) | SceneStyler · Cross-view guidance ablation | ArtFID: 22.809 -> 22.801 | No clear effect |
| train views in spatial order instead of random order (ablation) | SceneStyler · View order ablation | Long MEt3R: 0.2804 -> 0.2795 | Improved |
| optimize only color instead of all Gaussian parameters during stylization (ablation) | SceneStyler · Optimized parameters ablation | Structure Loss: 0.0607 -> 0.0318 | Improved |
| optimize only color instead of all parameters (ablation) | SceneStyler · Optimized parameters ablation | ArtFID: 23.215 -> 22.801 | Improved |
| raise capacity constraint tau from 0.01 to 1 (ablation) | CCFT loss · qualitative (single scene) | texture diversity | Improved |
| replace NNFM, FAST or Gram loss with CCFT loss (ablation) | SceneStyler · loss function ablation (qualitative) | style detail | Improved |
| use SceneStyler instead of ABC-GS in runtime | SceneStyler · Stylization runtime | minutes: 5.38 -> 8.35 minutes | Hurt |
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.
- StylizedGS — StylizedGS: Controllable Stylization for 3D Gaussian Splatting (TPAMI 2025) · better on 1 of 2 shared results
- SGSST — SGSST: Scaling Gaussian Splatting Style Transfer (CVPR 2025) · better on 1 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
- Multi-StyleGS: Stylized Gaussian Splatting with Multiple Styles (AAAI 2025 · similarity 0.94)
- Morpheus: Text-Driven 3D Gaussian Splat Shape and Color Stylization (CVPR 2025 · similarity 0.93)
- ReGS: Reference-based Controllable Scene Stylization with Gaussian Splatting (NeurIPS 2024 · similarity 0.92)
- Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles (NeurIPS 2025 · similarity 0.91)
- CLIPGaussian: Universal and Multimodal Style Transfer Based on Gaussian Splatting (NeurIPS 2025 · similarity 0.90)
- SGSST: Scaling Gaussian Splatting Style Transfer (CVPR 2025 · similarity 0.90)
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.
