If rendering an image in 2D, adding depth of field effects (blurring objects further from the focal distance) adds realism and draws the eye to the object of the image. With a 3D (i.e. stereo) image, looking at an object in the image at a given depth will makes objects at all other depths defocused (not blurred, but incorrectly aligned by the eyes, giving a double image). This means that if depth of field effects are used, there will be conflicting results: looking at an object that is at a different depth will cause that depth to be the only depth not having a double image, but it is also a depth that is blurred. This gives the object a property of being focused upon, and a property of not being focused upon. In a 3d still image, are depth of field effects detrimental to the acceptance of the image by the eye, or are there ways around this?
In traditional stereo 3D, I don't believe that there is a way to make a fixed focal plane feel natural to the viewer. When looking at an out-of-focus object in stereo 3D, the object remains out-of-focus, causing conflicting cues. The lens in the eye tries to adjust to bring the object into focus, but of course it won't succeed, causing eye strain and headaches.
However, there is hope outside stereo 3D: Lightfield displays, such as this nvidia prototype, go a different route. In stereo 3D, the light in the scene is already captured by two virtual (or physical) cameras, "baking" in the focal plane. Head-mounted displays like Oculus Rift then attempt to tape two displays in front of your eyes in such a way that the retina receives the exact same image that was captured by the camera. Lightfield displays go a different route: Instead of capturing two images ahead of time, they reproduce the entire 4D light field in front of your eyes, allowing your eyes to capture the image as if they were sitting directly inside the virtual scene. This has a number of benefits, including much smaller and lighter hardware as well as giving your eyes the ability to refocus.
If there is a way to make lightfield displays technically and commercially viable, then I believe they can remove the need for depth of field and fixed focal planes entirely and make VR feel a whole lot more comfortable for the viewer. However, it is likely not possible to construct lightfield screens, so televisions and cinemas won't be able to use this technology.
The authors of the article:
seemed to find that simulated depth of field effects were actually detrimental to the illusion of realism, while actual depth of field effects were beneficial.