A digital watermark embedder reduces a watermark signal as a function of time varying properties of video such that a watermark that would otherwise be static over frames is selectively reduced to make it imperceptible. The method computes a watermark signal corresponding to locations within a frame, where the watermark signal is mapped to locations in the video frame and is computed based upon attributes of the video within the frame. The method varies the strength of the watermark signal over time. The process of varying the strength includes reducing the strength of the watermark signal to make the digital watermark less perceptible in the video in locations where the video has time varying properties. The method embeds the watermark signal into the video at the locations with the varying strength.