Where, What, Why: Toward Explainable 3D-GS Watermarking

2026-03-09

AI Summary

This work proposes the first interpretable watermarking framework tailored to the native representation of 3D Gaussian Splatting (3D-GS), addressing the challenge of embedding robust yet imperceptible watermarks. The method employs a Trio-Experts module to select watermark carriers and introduces a Safety and Budget Aware Gate (SBAG) mechanism that dynamically allocates Gaussians for either watermark embedding or visual compensation, jointly optimizing robustness and fidelity under perturbation and bitrate constraints. By innovatively integrating channel-grouped masking to control gradient propagation with a decoupled fine-tuning strategy, the framework enables per-Gaussian attribution analysis, revealing the rationale behind watermark placement while effectively suppressing artifacts and preserving high-frequency details. Experiments demonstrate superior performance over state-of-the-art methods, achieving a 0.83 dB PSNR gain and a 1.24% improvement in bit accuracy, along with strong robustness, view consistency, and efficiency under various distortions such as compression and noise.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Safety and Robustness

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

As 3D Gaussian Splatting becomes the de facto representation for interactive 3D assets, robust yet imperceptible watermarking is critical. We present a representation-native framework that separates where to write from how to preserve quality. A Trio-Experts module operates directly on Gaussian primitives to derive priors for carrier selection, while a Safety and Budget Aware Gate (SBAG) allocates Gaussians to watermark carriers, optimized for bit resilience under perturbation and bitrate budgets, and to visual compensators that are insulated from watermark loss. To maintain fidelity, we introduce a channel-wise group mask that controls gradient propagation for carriers and compensators, thereby limiting Gaussian parameter updates, repairing local artifacts, and preserving high-frequency details without increasing runtime. Our design yields view-consistent watermark persistence and strong robustness against common image distortions such as compression and noise, while achieving a favorable robustness-quality trade-off compared with prior methods. In addition, decoupled finetuning provides per-Gaussian attributions that reveal where the message is carried and why those carriers are selected, enabling auditable explainability. Compared with state-of-the-art methods, our approach achieves a PSNR improvement of +0.83 dB and a bit-accuracy gain of +1.24%.

Problem

• 3D Gaussian Splatting
• watermarking
• robustness
• imperceptibility
• explainability

Innovation

• 3D Gaussian Splatting
• Explainable Watermarking
• Representation-Native
• Gradient Masking
• Decoupled Finetuning