Frame Rate, Not the Grid: A Confound in Single-Clip Optical ENF Video Forensics

Abstract

The optical electric network frequency, the faint mains-driven flicker imprinted on rolling-shutter video, is increasingly proposed for verifying where footage was captured and for flagging manipulated or generated video. This letter shows that reading the mains frequency from a single short clip is confounded by the camera frame rate: at the rates in common use, a harmonic of that rate lands in the same narrow band as the true flicker, so the recovered grid is set by the frame rate, not the supply, and the estimator returns a confident value instead of abstaining. The underlying aliasing is known; its forensic consequence is not, and is the contribution here. On the 804 clips with a measurable in-band line the flicker feature does not separate genuine from generated video (area under the curve 0.509), while the recovered grid tracks the frame rate exactly for the generators, and a controlled synthesis confirms the mechanism against ground truth for genuine grids. Single-clip grid, and hence region, identification is therefore unreliable at the standard rates, and a comb-aware estimator that models the frame-rate harmonics restores the distinction between a genuine grid and a frame-rate comb under controlled conditions.

4 pages, 2 figures, 2 tables. Submitted to IEEE Signal Processing Letters. Code and data: https://github.com/RyanSolent95/enf-video-forensics