Computer Graphics Photography Dan Anderson This has been tried for the Nikon FM2, with Nikon 55mm f/2.8 macro lens. It should also work the same for other cameras/lenses near 50mm (but see the last detail, below). You have to use a shutter release cable and a tripod. The first detail of the details list is not a detail!!! The exposures stated here are based on trying it and looking at the results. Different graphics softwares with the same screen seem to have different brightnesses, and therefore different exposures than what it says here. Film: Kodak "Elite" 100 slide film. If you need near-perfect reproducibility, use Kodak "Lumiere" 100 professional slide film. Marie Frodo electron density, etc., displayed on an Evans and Sutherland screen: Exposures are 5 or 6 seconds at f/5.6 . The camera's shutter speed dial should set for B, the ring on the lens should be set for 5.6 . Use the blinking cursor of an adjacent terminal to count seconds. Silicon Graphics O CPK: So far, 3 seconds at f/8 seems good. Be very careful with the Z translation. The atoms get real dark when they are moved back. Next time, we will pay closer attention to the location of the "slab". Silicon Graphics Insight CPK: Looks like 1/2 second at f/5.6 (that's "2" on shutter speed dial). Silicon Graphics Insight worm-like critters: 2 seconds at f/5.6 seems about right, but 3 sec is way too much. Details: Frame 1 or 2 should define the left and right edges of the frame. Take a picture looking out the window, or 1/4 second exposure of the graphics screen with the room lights on. The automatic frame cutter at the processing lab will extrapolate from there. The automatic frame cutter cannot be properly adjusted to an almost completely blank roll of film by an uninitiated photo lab worker. Frame 1 should be something recognizable, otherwise they might ignore it and cut the film in a near random place. To minimize the probability of massacred slides, write a special instruction on the processing envelope: "Set automatic film cutter on first bright frame." The optimum exposure for Frodo on an Evans and Sutherland screen is 5.5 seconds at f/5.6 . I synchronize my big toe to the blinking cursor of an adjacent terminal. The down-toes are integer seconds; the up-toes are 1/2 seconds. The first down-toe is 0 seconds. We oughta have a metronome. Set up the camera on a tripod, perpendicular to the center of the screen. The screen perpendicular is not horizontal. Align the camera to the edges of the screen or to a geometric test pattern. Look for reflections on the screen from the point of view of the camera, not from a standing viewpoint. When Frodo first starts, shut off the mid-screen characters with the TERMINAL button at top center, and turn on the function keys, so that the menu and axes displays may be blacked out. In the DISPLAY menu, Push the KBD OFF button to get rid of the lower left corner characters. The stereo effect should be on, but the "viewing angle" should be 0 degrees (the default is 3 degrees). To change view angle, push 5 on the terminal, set view angl to 0 with the dial, then press 1 to get back to the view dials. I adjust for critical focus using the focus magnifier attached to the camera. This results in very good focus for the middle of the picture. In principle, the lens should be focused off-center to achieve "hyperfocus" that includes the entire curved screen. It has worked anyway because Frodo's display region is square. Press the shutter release cable just enough to open the shutter. More force will eventually destroy the shutter mechanism. The shutter release cables that I buy for the lab have "B" and "T" modes. The camera has only B mode. The mode for the cable is set by a collar on the part of the cable that goes in your hand. "B" means rubber bulb: As long as you squeeze the (symbolic) rubber bulb, the shutter stays open. "T" stands for time: You push the button, then Time T later push the collar. I always use "T" mode for graphics. Estimating the exposures with the light meter is complicated. The screen displays colored dots on a black background. The light meter needs to look at a solid block of color, large enough to fill the metering region. The exposure can be estimated from the relative brightness of a window and the thing to actually be photographed. Even if the window completely fills the meter, the meter is still averaging the dots with the black grid of the screen, while the thing to be photographed is the dots, not the grid. Although windows look like about 2 stops brighter than the atomic displays, the required exposures have had to be about 1/2 stop more than what I would estimate from the meter. The measurement is further complicated by the interaction of screen flicker, meter averaging time, and film reciprocity. Fast light meters, like mine, completely fail for this application because they track the screen flicker up and down. Reciprocity characteristics are a function both of overall exposure time, and of the point-by-point history of how fast the photons arrive. Empirical recipe: First, focus the camera as though you are going to take the real picture. Without changing the focus, fill the light meter region with a solid window. In many Nikons, the light meter region is signified by the black circle in the viewfinder. Meter that as though it is the thing of interest. Dial in 2 stops more exposure than indicated by the meter as a minimum exposure. For example, if the light meter indicates 0.5 second at f/5.6, then increase the exposure to 2 seconds, then try to get 2, 2.5, and 3 seconds by counting cursor blinks. Is computer graphics photography even a photographic problem? A black on white laser printer output from a postscript file might communicate to the reader better than a colored lines on black background display. Most scientific literature is read as photocopies, and therefore what the reader sees is not the screen, the photograph, the print, or the journal page, but the photocopy resulting at the end of that process. Think through the entire chain when preparing publications. The Nikon 55mm macro lens seems to be the shortest focal length that does not give horrible distortion due to the curvature of the screen. If you display a box around your molecule, it will still be curved with a 55mm lens. Ideally, use a macro lens around 100mm to minimize distortion. On the other hand, what they don't know won't hurt them. I would worry more about the 20-30% overall squeeze in one direction that sometimes seems to be typical... I'm not sure if the exposure recommendations transfer to other lenses because f stops are theoretically different than t stops. For instance, I don't think the Micro Nikkor 55mm lens transmits as much light as my Olympus lenses at the same f stop.