StegaNeRF achieves reliable recovery of hidden information with minimal impact on the NeRF rendering quality. This work offers a promising outlook on ownership identification in NeRF and calls for more attention and effort on related problems.
Recent advances in neural rendering imply a future of widespread visual data distributions through sharing NeRF model weights.
However, while common visual data (images and videos) have standard approaches to embed ownership or copyright information explicitly or subtly, the problem remains unexplored for the emerging NeRF format.
We present StegaNeRF, a method for steganographic information embedding in NeRF renderings.
We design an optimization framework allowing accurate hidden information extractions from images rendered by NeRF, while preserving its original visual quality.
We perform experimental evaluations of our method under several potential deployment scenarios, and we further discuss the insights discovered through our analysis.
StegaNeRF signifies an initial exploration into the novel problem of instilling customizable, imperceptible, and recoverable information to NeRF renderings, with minimal impact to rendered images.
TL;DR: The first framework to instill customizable, imperceptible, and recoverable information to NeRF renderings.
Left: Rendering produced by StegaNeRF. Right: Recovered hidden image from the corresponding rendered frame.
Applying StegaNeRF on single-scene NeRF training.
The hidden information are embedded in the rendering and can be revealed through visualizing the residual error against the rendering produced by a standard NeRF without steganographic training. Across all the rendering viewpoints, the same hidden information is robustly recovered, with only very slight variations.
Applying StegaNeRF on multiple scenes at once.
Applying StegaNeRF to embed multi-modal information. Please see the video below for the detailed exhibition.
Recovered Audio + Text