MultiNeRF: Multiple Watermark Embedding for Neural Radiance Fields

MultiNeRF Teaser

MultiNeRF embeds multiple watermarks in NeRF (TensoRF) at training time. Watermarks are keyed by unique IDs at rendering time, allowing independent watermark activation.

Abstract

We present MultiNeRF, a novel 3D watermarking method that enables embedding multiple uniquely keyed watermarks within images rendered by a single Neural Radiance Field (NeRF) model while maintaining high visual quality. Our approach extends TensoRF by incorporating a dedicated watermark grid alongside geometry and appearance grids, ensuring higher watermark capacity without entanglement with scene content. We propose a FiLM-based conditional modulation mechanism that dynamically activates watermarks based on input identifiers, allowing multiple independent watermarks to be embedded and extracted without retraining. Experiments on NeRF-Synthetic and LLFF demonstrate improved robustness and watermark capacity while preserving rendering quality. MultiNeRF generalizes prior single-watermark methods, providing a scalable solution for securing ownership and attribution in 3D content.

MultiNeRF training pipeline: encoder, watermark embeddings, differentiable noise, perceptual + patch losses.

Performance comparison for single watermark embedding.

Bit accuracy across SYN and LLFF datasets for multiple watermark IDs.

Poster

BibTeX

@inproceedings{Kulthe:MultiNeRF:ICLRWS:2025,
  author = {Yash Kulthe and Andrew Gilbert and John Collomosse},
  title = {MultiNeRF: Multiple Watermark Embedding for Neural Radiance Fields},
  booktitle = {International Conference on Learning Representations (ICLR'25) - 1st Workshop on GenAI Watermarking},
  year = {2025}
}