SMPTE Report: Does 3D Need To Adjust for Different Screen Sizes?

When it comes to monetizing 3D sports for TV, one of the best bets is believed to be distribution in movie theaters, on laptops, and even on handheld devices to ensure as many available revenue channels as possible. But a primary issue facing that model is that, as 3D content is scaled up or down to fit a larger or smaller screen, there is an impact on the perceptual effects of 3D. At this week’s SMPTE Conference in Los Angeles, Clive Henry Gillard of Sony in the UK described the significance of that impact and whether it can cause issues for viewers.

“With 2D, scaling everything just works, but, in 3D, the problem is, of course, the third dimension,” he explained. “While the horizontal and vertical size changes, the depth is not scaled. But, if you scale the depth, the object will seem miniaturized.”

And then there is the pain factor. There are industry production professionals calling for positive parallax to be 1% of screen size. But, when the screen is larger than the one the content was originally created for, that parallax will move beyond 1%, causing the eyes to diverge and, potentially, leading to pain. And when the screen is smaller, vergence issues become the primary concern.

In a movie theater, for example, the viewer is sitting farther from the screen, and the eye muscles are relaxed, allowing the content creator to have depth that is 1% behind the screen and 2% in front without negative impact.

“But, with TV, the viewer is looking at the screen more closely, and the eye muscles are more active,” Gillard noted, “so the 3D depth can only extend to 1% in front of the screen.”

He added that most 3D content also includes 2D cues to depth, helping decrease the negative effects of depth distortion.

“Depth distortion is not a significant issue, as 2D perspective cues will help the viewer,” he explained. “And horizontal image translation, a method where the left and right images are shifted with respect to each other to push the object to where it was originally, results in no depth distortion.”

Gillard discussed a recent Sony study in which two groups of viewers looked at images created for an 8-meter screen on a .92-meter screen. They also sat 1.38 times picture height from the screen. The goal was to see whether the depth scaled linearly relative to the decrease in screen size or scaled non-linearly and, therefore, without the ability to predict the perceived depth.

In the study, Gillard said, one group had content that was scaled down but without perspective cues, while the other group had content scaled down with perspective cues, such as other objects that were smaller or larger and gave a sense of relative depth.

“The picture with the 2D perspective clues,” he added, “was much easier for everyone to work out where the object was, and, on average, viewers were able to place the column closer to the linear depth than the distorted depth.”