About 12 years ago or so, I helped on the translation of a research project on this issue (the camera / display one – not the how much of life is Star trek ) by a Japanese company. The details have long since left the old grey matter, but they were looking into several options. One potential solution involved having the camera behind the screen and rapidly alternate between transparency to allow the camera to accept light in and luminance to display the picture. The paper speculated that if done correctly, this would be an optimum solution. However, no actual research was being conducted at that time.
They pointed out that there is a limited range in degrees which people sense that the other person is “looking” directly at them. Putting a single camera next to (on the top, side or bottom of) the display won’t cut it.
Technology has changed in the last 12 years, but this is still not an easy problem.
LED screens, as suggested by **scr4 ** are problematic. The most obvious is having a low enough resolution to permit large enough holes would render the image unusable for VC purposes.
There are several problems with a pin-hole camera embedded within a display. The smaller the diameter of the camera, the lower its resolution. If the idea is to have a more natural feel for VCs, there is a limit to how small the lens can be. Practically, there would also be major manufacturing issues, with the whole thing having to be redesigned around the camera unit. Since something like this would only be used for VCs – no one wants to buy a 60" plasma with a hole in the center for their home theater – then sales would be limited. Currently there is enough of a market for this.
Software rendering also isn’t practical now. Even if it could be done, the cost and complexities of the workstations rule this out.