Reproduce “Tune-Your-Style: Intensity-tunable 3D Style Transfer with Gaussian Splatting” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yian Zhao, Rushi Ye, Ruochong Zheng, Zesen Cheng, Chaoran Feng, Jiashu Yang, Pengchong Qiao, Chang Liu and 1 more
Researchers from Peking University and collaborators introduce Tune-Your-Style, a method for intensity-tunable 3D style transfer utilizing Gaussian Splatting. The approach allows users to continuously adjust the artistic style's intensity, mitigating issues of over- or under-stylization, and produces visually superior, multi-view consistent results compared to existing 3D stylization techniques.
Key insights
- Explicitly modeling style intensity through learnable attribute offsets for 3D Gaussian primitives, modulated by a 'style tuner', provides continuous and granular control over stylization.
- Combining 2D diffusion priors with a cross-view style alignment strategy, including anchor view feature injection and content calibration, is effective for achieving multi-view consistent 3D stylization from varying perspectives.
- A two-stage optimization approach is critical for the stable and effective learning of tunable stylization, first establishing a full stylization baseline and then progressively learning intermediate intensity levels.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.