A psychophysical study isolating judder

Dolby Laboratories' Scott Daly presented a paper on "A psychophysical study isolating judder using fundamental signals," co-authored by Ning Xu and James Crenshaw (Dolby Laboratories); and Vickrant J Zunjarrao (Microsoft). Movie content displayed at different frame rates has been observed and described for a number of years. "While the terms are not entirely solidified across the industry, there are four main degradations of the signal as compared to unsampled, or real-world, motion," said Daly. He enumerated those as non-smooth motion (most often referred to as judder, or strobing); the appearance of false multiple edges; flickering (a counter-phase spatiotemporal frequency along moving edges); and motion blur In natural imagery. "Judder is an umbrella term describing several visible attributes: non-smooth motion, false multiple edges, edge flickering and motion blur," he said, demonstrating these four attributes.

He spoke about the once-common use film look vs. video look. "It ultimately became a matter of frame rate," he said. "Theories of preference for cinema rate rate include that 24ps motion blur is a tool for directing attention as well as hiding bluescreen seams and fake bears. Other beliefs are that cinema should not look real because that hinders imagination is another belief; people associate judder with cinema and smooth motion with video. "Preferences for film look is an appearance question," he said. "Douglas Trumbull's preferences for judder depend on an object's motion. His current system allows variable frame rates within a scene; and the shutter angle can be varied somewhat as well to control motion blur."

The goal of the study that Daly presented was primarily to investigate the non-smooth motion attribute. "We avoid biasing viewers by not telling them what to look for," he said. He described the test methodology including the signal display system and viewer/viewing conditions. He showed images of the fundamental test signals shown to the observers. Some key findings were that frame rate was the primary variable affecting judder, both with simple and complex imagery. With motion speed, judder increases with increased speed and contrast had a strong effect on judder although brightness didn't give the expected result. When contrast is allowed to change as a result of brightneses, judderness increases with increasing brightness with statistical significance. The shutter angle had little affect on judder when angle step size was small; there is decreasing judder with increasing shutter angle. With regard to direction of motion, horizontal has more judder than vertical.

The test shows qualitative agreement with all visual psychophysics data and models for detection, except Ferry-Porter law, and qualitative agreement with the Cinematographer's Handbook. Future work will include comparisons at higher frame rate and other tests with other stimuli to isolate edge flicker and multiple edges.