Although Nintendo has done a great job at introducing glasses-less 3D technology to a mass audience, the Nintendo 3DS display has a couple of things that would need serious improvements: brightness and power consumption. The design of the 3DS display is remarkable because it uses two LCD displays stacked on top of one another and a “parallax mask” makes it so that you eyes see different images, giving the impression of depth. Blocking the mask lets users switch back to a 2D mode. However it also causes the image to be less bright than regular 2D LCDs because the light coming from the back LCD has more layers to go through. To compensate for that, Nintendo has to crank up the light of the back LCD, which causes the battery to deplete faster (3hrs or so of battery life, versus 6 for the DS).
A Team of MIT researchers (Associate Professor Ramesh Raskar, postdoc Douglas Lanman and graduate student Matthew Hirsch) have developed HR3D, a 3D glasses-less display technology that does not have the brightness weakness, or power requirement issue of the current Nintendo 3DS display.
It has great potential for handhelds, but also for televisions, because current commercial glasses-less technologies tend to have poor viewing angle, making them more suitable for a 1-person device like the 3DS, than a group-device like a television.
HR3D is smart and works by adapting the mask between the two LCD displays in real-time so that they are optimum for each image or animation frame. It’s like having a custom mask for each frame displayed. The result is a much better view angle and it doesn’t require as much backlight power as current technologies.
The obvious downside is that as the image on the screen gets animated, the whole mask needs to be recomputed -in realtime-, and that… can be very costly. The MIT team has recognized that if they spent all that power savings into expensive mask computations, things aren’t as good as they should be.
But this is still a great invention. For one, many devices don’t have strict power requirements. A television for instance, can afford to spend more power in mask computation, where a handheld cannot. Secondly, the mask computation seems to be image-based (2D) and the performance requirement might only on the resolution of the image, so it seems like a “fixed” processing cost at any given resolution. As processors get faster, the relative computing cost goes down.
Playback a simple .mp3 music file used to be hugely computationally expensive on PCs but these days, it barely takes 1% of the resources at any given time. The same thing is probably true for HR3D. At some point, the mask computing cost should become negligible.