FlyMark: Training-Free Invisible Watermarking of 3D Gaussian Splatting via a Fruit Fly Connectome

2026-10-08

AI Summary

This study addresses the vulnerability of existing 3D Gaussian Splatting (3DGS) ownership watermarking methods, which typically rely on training procedures or dedicated decoders and are thus prone to failure when such tools become unavailable. To overcome this limitation, this work proposes a training-free, deterministic parameter-level watermarking framework. Specifically, the method leverages the *Drosophila* visual connectome to generate carrier directions and integrates Quantization Index Modulation (QIM) encoding with sparse least-squares optimization for subtle color refinement, thereby enabling both watermark embedding and extraction without requiring a learned decoder. By preserving geometric structures and high-order appearance properties, the proposed approach achieves high bit accuracy, excellent visual fidelity, and strong robustness against various attacks, offering a resilient new paradigm for 3DGS copyright protection.

Abstract

A trained 3D Gaussian Splatting (3DGS) scene ships as a portable parameter array that can be copied, pruned, requantized, or repackaged outside its training pipeline, so ownership evidence is most useful when it lives in the released parameters and remains checkable long after the embedding tooling is gone. Existing 3DGS watermarks typically tie embedding or extraction to scene optimization, a learned decoder, or rendered views, so the evidence survives only as long as a second trained artifact does. FlyMark instead writes a keyed message into the parameters a 3DGS file already stores. Its carrier directions are derived from the photoreceptors of a published connectome, a citable versioned artifact that fixes the geometry exhaustively and leaves nothing to tune per scene. A virtual observer reads cone-wise apparent luminance along a scene-normalized orbit from stored centers, colors, and opacities; a keyed dithered quantization-index-modulation code replicates each message bit across these observations; and one sparse bounded least-squares solve realizes the targets through achromatic shifts of existing degree-zero colors under a hard per-channel linear-RGB bound. All geometry and higher-order appearance parameters are preserved bit-identically, and extraction needs only cone queries, rounding, and majority voting. Under a model-domain threat model on synthetic and real scenes, FlyMark attains high clean bit accuracy and visual fidelity while cleanly separating matched from wrong keys.

Problem

• 3D Gaussian Splatting
• Invisible Watermarking
• Training-Free
• Ownership Protection
• Intellectual Property

Innovation

• Training-Free Watermarking
• 3D Gaussian Splatting
• Fruit Fly Connectome
• Quantization Index Modulation
• Bounded Least-Squares