Altering the visual impression of 3D Gaussian splatting via style transfer
Abstract
In recent years, 3D Gaussian Splatting, which uses neural networks to generate 3D scenes from multiple images, has been attracting attention. Furthermore, by applying style transfer to the generated 3D scenes, it is possible to generate more impressive 3D scenes. However, when applying style transfer to the entire 3D scene, noise in areas such as the background is overly emphasized. Therefore, our method uses segmentation to apply style transfer only to the target area, thereby generating more impressive 3D scenes.
Fields of Study & Subjects
3D Shape Modeling and Analysis
Generative Adversarial Networks and Image Synthesis
Computer Graphics and Visualization Techniques
Publication details
- Journal / source
- International Workshop on Advanced Imaging Technology (IWAIT)
- Pages
- 126–126
- Publication date
- 27 February 2026
- Language
- EN