Moiré Video Authentication: A Physical Signature Against AI Video Generation

Dimensions overall score 7.0

• The Moiré motion invariant shows that fringe phase and grating image displacement are linearly coupled by optical geometry, independent of viewing distance and grating structure.technical 90%
• Real cameras produce Moiré effect signatures naturally that generative models cannot faithfully reproduce.technical 85%
• Real and AI-generated videos produce significantly different correlation signatures when tested with the Moiré motion invariant.result 90%
• The Moiré effect can be exploited to create a robust means of differentiating real from AI-generated video.method 80%
• Deterministic optical phenomena can serve as physically grounded, verifiable signatures against AI-generated video.technical 85%
• The verification method involves extracting both fringe phase and grating image displacement signals from video and testing their correlation.method 90%
• The approach uses a compact two-layer grating structure to create the Moiré interference fringes.method 85%