3D-GSW: 3D Gaussian Splatting for Robust Watermarking

TL;DR

The paper addresses the need to protect 3D-GS content by embedding robust watermarks into rendered views. It introduces Frequency-Guided Densification (FGD) to prune non-critical Gaussians and split high-frequency components, a gradient mask to preserve rendering quality, and a wavelet-subband loss to strengthen high-frequency fidelity, all guided by a pre-trained watermark decoder. The approach achieves strong invisibility and robustness against image distortions and model alterations, outperforming state-of-the-art baselines while maintaining real-time rendering performance. This work significantly advances ownership protection for 3D-GS assets in rapidly evolving 3D content ecosystems, with potential for multi-modal data embeddings in future work.

Abstract

As 3D Gaussian Splatting (3D-GS) gains significant attention and its commercial usage increases, the need for watermarking technologies to prevent unauthorized use of the 3D-GS models and rendered images has become increasingly important. In this paper, we introduce a robust watermarking method for 3D-GS that secures copyright of both the model and its rendered images. Our proposed method remains robust against distortions in rendered images and model attacks while maintaining high rendering quality. To achieve these objectives, we present Frequency-Guided Densification (FGD), which removes 3D Gaussians based on their contribution to rendering quality, enhancing real-time rendering and the robustness of the message. FGD utilizes Discrete Fourier Transform to split 3D Gaussians in high-frequency areas, improving rendering quality. Furthermore, we employ a gradient mask for 3D Gaussians and design a wavelet-subband loss to enhance rendering quality. Our experiments show that our method embeds the message in the rendered images invisibly and robustly against various attacks, including model distortion. Our method achieves superior performance in both rendering quality and watermark robustness while improving real-time rendering efficiency. Project page: https://kuai-lab.github.io/cvpr20253dgsw/

Figures (27)

• Figure 1: The unauthorized use of the 3D Gaussian Splatting model. Our method ensures that the watermark remains detectable even in distorted images and under model attacks.
• Figure 2: 3D-GSW Overview. Before fine-tuning 3D-GS, Frequency-Guided Densification (FGD) removes 3D Gaussians based on their contribution to the rendering quality and splits 3D Gaussians in high-frequency areas into smaller ones. We also construct a gradient mask based on the parameters of an FGD-processed 3D-GS. During the fine-tuning, we apply the Discrete Wavelet Transform (DWT) to the rendered image for robustness, using the low frequency as input to a pre-trained message decoder. For rendering quality, we design a wavelet-subbands loss that utilizes only high-frequency subbands. Finally, 3D-GS is optimized through the $\mathcal{L}_{total}$.
• Figure 3: Rendering quality comparison of each baseline with our method. We doubled the scale of the difference map. Our method outperforms others in bit accuracy and rendering quality, using 32-bit messages for the qualitative results.
• Figure 4: We present a rendering quality comparison for 32-bit, 48-bit, and 64-bit messages. The differences ($\times$2) between the watermarked image and the original image. Since manipulated areas are high-frequency areas where the people's eyes are less sensitive, the rendered image with our method looks more realistic and natural.
• Figure 5: Rendering quality comparisons with full method(ours), without FGD, without gradient mask, without wavelet-subband loss, and base model. All images have 32-bits embedded.