Light2Lie: Detecting Deepfake Images Using Physical Reflectance Laws
Kavita Kumari
Network and Distributed System Security (NDSS) Symposium 2026 · Day 2 · Multimedia Forensics
This talk presents **Light2Lie**, a deepfake detection approach based on the insight that real images follow physical light reflectance laws while AI-generated images do not. By modeling each pixel as a **microfacet** using the **Blinn microfacet theory** from computer graphics, the system computes a **specular reflection score** for each image that captures how light interacts with the surface. Real images exhibit complex, highly variable reflectance patterns because they result from actual light-surface interactions, while AI-generated images produce smoother, less variable patterns because generators are not trained on these physical laws.
AI review
An interesting idea -- using physical reflectance laws to detect deepfakes -- but the execution feels preliminary. The core assumption that AI images are smoother than real ones is a strong simplification that may not hold as generators improve. The base reflectivity assignment (0.96 for fake, 0.04 for real) feels circular, and the speaker acknowledges the results are 'not perfect' with known failure modes in extreme lighting. The approach outperforms weak baselines on one dataset but needs more rigorous adversarial evaluation.