Helvensteijn.com

First of all, I want to make it clear that, while I have done a bit of research, I am by no means an expert on this matter, I just find it very interesting. This article is purely about my personal experience with 3D displays, and what I think is wrong with them, based on that experience.

Having said that, let’s clarify what I mean by “3D displays”. I’m referring to the stereoscopic kind, these flat displays projecting a different image at each eye. For this article, it’s not relevant how they accomplish that. What’s relevant is that they rely on stereopsis, meaning the disparity between what each eye sees, which gives us a cue of relative depth.

So, what exactly do I think is wrong with them? It boils down to two things: convergence and motion parallax. Both are fixed, which your brain doesn’t expect. There are other depth cues we see in everyday life that are still missing from 3D displays, but I think these two are most important.

To start off with convergence. This is the fact that when both eyes focus on the same object, they turn towards each other. The amount of convergence depends on the distance to the object. The closer the object, the more the eyes will turn inward. With a stereoscopic display, there really is only one distance, which is the distance to the display itself. To see things sharply, both eyes must focus, and thus converge, on the display. This fixed convergence conflicts with the stereoscopic depth cue that the display provides.

Then, there’s motion parallax. When you look at an object and shift your head, you should see the object from a different perspective. In stereoscopic displays, depending on the technology used, when you shift your head either nothing happens — again conflicting with the depth illusion the display is trying to evoke — or the 3D effect falls apart completely, which is even worse.

The parallax problem has, in fact, already been solved, at least with generated imagery (i.e. video games, computer animations). I’ve seen demonstrations of 2D displays creating an illusion of depth, by changing the perspective using head tracking. Such a system could relatively easily be adapted for 3D displays, at least those using special glasses. Auto-stereoscopic displays (those without special glasses) generally have a parallax barrier or lenticular lenses fixed to the display, so dynamically shifting perspective seems somewhat more complicated there, unless the parallax barrier or lenticular lenses themselves could be shifted.

In 3D films, motion parallax would probably require more than two cameras, or maybe two widely spaced cameras with a computer interpolating the in-between perspectives. In either case (3D films or generated imagery), two fixed perspectives (one for each eye) is one thing, providing a whole range of perspectives might not even be feasible for most applications.

As for convergence, I guess it would be a lot more difficult, if not impossible, to solve. The way I see it, as long as the display itself is flat, convergence will always remain a problem.

Then, there’s a third problem which I’ve observed on my Nintendo 3DS. No matter if an object is in the foreground or in the background or in between, everything is sharp. It’s like it has an infinite depth of field. While the stereoscopic effect in and of itself is quite convincing, it makes it feel somewhat unnatural to me. 3D movies are better in that respect. The foreground is in focus with the background blurry, or vice versa. Perhaps there are games that are more like 3D movies in that regard, but both the OS and the games I’ve played so far exhibit this “infinite depth of field” effect.

In any case, while personally, I don’t have much problems looking at 3D displays, I can definitely see how they can cause headaches for some. They give conflicting depth cues, and as with every other conflicting situation, some people cope better than others.