Using shutter glasses for 3D molecular graphics

Background:

One wouldn't think that using shutter glasses primarily designed for gaming can yield good results for the scientific community. However, that is not the case. It is possible to use $55 shutter glasses with a $100 video card with any modern PC as a 3D modeling workstation.

Let's start with the shutter glasses -- the product that makes all of this possible. The 3D Revelator glasses, available from www.elsa.com, use small LCD screens to alternatively block out the left and right eye, allowing the two eyes to see different images, thereby giving the illusion of 3D objects on a 2D screen. This technology has existed previously, but there are two fundamental distinctions in this incarnation. First and foremost, these shutter glasses work with *any* program that utilizes the Direct3D API to draw the 3D molecule, and most programs that use the OpenGL API. Previously, a program had to have special support for 3D glasses, which would always limit the availability of software. With these glasses, almost any program that uses Direct3D or OpenGL, whether it be for displaying molecules or gaming, can be given a realistic 3D appearance. The drivers for the shutter glasses work as follows: The scene described by the program to the graphics API (Direct3D or OpenGL) is rendered from two different perspectives, one for each eye. The alternate images are then very quickly flashed one after the other, and the eye that is not supposed to see one of the images is blocked out by the shutter glasses. This brings us to the second fundamental distinction between this incarnation and previous ones -- there is no flicker. Previously, the two images, each intended for one eye, did not refresh very often, which resulted in distracting flicker of the model. These glasses, however, refresh at up to 70Hz for each eye, or a total rate of 140Hz. This is fast enough to make any flicker disappear completely. In one word, the images look good.

These glasses do require a certain subset of video cards in order to function. The only video card limitation imposed is that the card has to carry a TNT, TNT2, GeForce, or GeForce2 processor (in order of speed, listed lowest to highest). Cards that carry these processors are widely available for various price ranges. I have an older Diamond Viper v550 that uses a TNT chip, and these glasses work it. I also have the newer Elsa Gladiac that uses a GeForce2 chip, and the glasses work with that as well. The GeForce2 will render the molecule faster than the TNT, allowing for smoother rotation, but in the end both setups work the same for displaying a stationary 3D model. Obviously the newer chips are going to be faster and more expensive, while the older chips will be slower but cheaper. My Elsa Gladiac does a very adequate job of rotating molecules with hundreds of atoms, and its current sale price, direct from Elsa, is $329. My Diamond Viper v550, which is no longer available from Diamond but available from other resellers goes for $60 to $100. Both work. Any card that carries one of the aforementioned chips will work, so a card from Elsa is not required.

I have found that the monitor used must support at least a 120Hz vertical refresh rate to get good results; most modern monitors fit this criterion.

Results:

I have gotten 2 molecular visualization programs to work with the Revelator shutter glasses. One of them is Swiss-PdbViewer, and the other is jvms. The former uses straight OpenGL, while the latter uses the Java language and its Java3D API, which can use either OpenGL or Direct3D. With my GeForce2, the shutter glasses only work through Direct3D. I heard rumors (soon to be tested -- one of my friends has a TNT2) that they will work directly with both OpenGL and Direct3D with a TNT2 graphics chip. There is a wrapper for OpenGL that will use Direct3D to render the images by SciTech Soft called GLDirect. It allows programs (like swiss-pdb viewer) that use OpenGL to work with the shutter glasses. MolMol, unfortunately, crashes while starting up when the SciTech wrapper is activated.

Overall, molecules look quite striking. After the eyes fully adjust (takes about 10 seconds for me), it is very easy to see what portions of a protein are in front, and which are in the back. Rotation of more complex molecules becomes slow (about 3 frames per second), but that is expected of a non-industrial-strength graphics setup. 

HOWTO:

Here's how to make it all work:

My Setup:

Here is some pertinent information about my system setup:

 


DISCLAIMER:

I take no responsibility for the validity and/or correctness of the information provided above. I don't work for or get any rewards or incentives from any of the mentioned companies. All trademarks are the property of their respective owners. If you have problems, please do not come to me. Instead, deal with your hardware/software providers.

Written by Sergey Klibanov, 08/04/2000