- Qualitative Comparisons: We compare our method against other state-of-the-art (SOTA) approaches: StyleGaussian (Liu et al., 2024), IGS2GS (IGS2GS Reference), and Ref-NPR (Zhang et al., 2023) on three datasets (Chair, Hotdog, and Mic) using three styles: Cartoon, Sky Painting, and Fire. + Qualitative Comparisons: We compare our method against other state-of-the-art (SOTA) approaches: StyleGaussian (Liu et al., 2024), IGS2GS (IGS2GS Reference), and Ref-NPR (Zhang et al., 2023) on three datasets (Chair, Hotdog, and Mic) using three styles: Cartoon, Sky Painting, and Fire.
The horizontal axis represents the compared methods, while the vertical axis displays different datasets. Our method effectively preserves the original model details, including: