|
SHOOT YOUR MONITOR: PROJECTION SLIDES THE FAST AND EASY WAY The following contains material presented at a 2002 National Stereographic Association meeting workshop. It describes how to make reasonable-quality 35mm slides for projection on the cheap by taking pictures of your computer monitor.
SHOOT YOUR MONITOR Prof. John Hart Program in Atmospheric and Oceanic Sciences University of Colorado Boulder, CO 80302 February 2002 nimbus.colorado.edu/hart/science.htm QUICK START SUMMARY I make good quality slides for projection from digital files by shooting my monitor with an SLR. When viewed under magnification these slides are just as good in resolution as those obtained from a professional lab using an MGI Solitaire film recorder (4K), and are only slightly poorer than those gotten from a LaserGraphics 16K film recorder (a very expensive machine) and costing $3 each (versus 30 cents for the monitor shoot). The colors of my slides ARE BETTER (although you probably could tweak the film recorder, at increased cost), and the contrast is good too. HERE IS MY LATEST (FEB 2004) TECHNIQUE: Mitsubishi Diamond Pro 2060U monitor set to 2048 x 1556 native resolution. Contrast 30%, Brightness 80%, Color Temp 5000K. Pentax SLR (with 50 mm lens) mounted appropriately: basically square and frame filling after accounting for viewfinder coverage (see below). Provia 100F high grain positive slide film shot at ISO100, processed normally, and exposed for 2 seconds at f 8. Put all the files (2048 x 1556 high quality jpegs) in a directory. Open one file with "Windows Picture and Fax Viewer". Hit F11. This puts you into the full frame slide show mode. Every 5 seconds the next file is displayed. Just sit there and bang off a 2 second exposure for each image when it pops up. You can make 36 slides from digital files in about 3 minutes!!! THE DETAILS: How effective is the simple method of making projection slides by photographing your computer screen with a film camera? What methods can be used to control exposure, contrast, and color? How do the results compare with other methods? Problem: You have digital images and want to project them. 2 Slide Projectors: ~ $700 to purchase a pair. This gives you a resolution of ~ 2000 lines (or 4000 pixels) across the screen, and ~ 1500 lumens of light output ("high-bright module", f 2.4 lenses...). 2 Digital Projectors: ~ $14000 to purchase a pair having a resolution of 600 lines or 1280 pixels (horizontally) at high contrast across the screen. Brighter, with ANSI ~ 2000 lumens, or much higher if need be (and no cooked slides....). For the time being, converting digital images to slides is cost effective and useful in situations where digital projection facilities are unavailable (e.g. at scientific meetings in foreign countries). If motion capabilities and/or high brightness are required then digital projection may be necessary – at a sacrifice of resolution. ___________________________________________________
Available Methods for making slides from Digital Files: "Film Recorders" (CRT) - to own you pay ~$4K, or $2 - $5 / slide at a photo laboratory. Examples: MGI Sapphire, MGI Solitaire, Lasergraphics LFR, Polaroid Pro Palette. These are rated in "addressable pixels" across the screen - 4K, 8K, etc. There is much more involved than just addressing the CRT screen. 8K does not mean you will see 8000 pixels on a piece of film!! "Laserjet2000" (1500 line RGB laser film recorder) – to own ~ big big bucks, ~ $6 per slide at the local lab (if they have one of these fancy machines). Continuous tone laser beam writing on large sheets of film at 2000 dpi. "Shoot your monitor". 1024 line or 2048 pixels across. Costs ~ $500 if you don’t have such a high resolution computer monitor. But operating cost is only 25 cents or so per slide. Can simply photographing a high resolution monitor with your old SLR produce decent slides that still exceed the resolution of current digital projectors? __________________________________________________ Proper mechanical setup and Photoshop corrections are required to shoot your monitor:
CONCLUSIONS Monitor shooting (using 2048 x 1556 CRT) is effective for digital source files not exceeding this resolution (i.e. cameras not much > than 5MP). The resolution so obtained exceeds that of current digital projectors by a significant amount. Slide quality is good and on most slide projection systems it is not possible to see the pixels of a 2048 x 1556 monitor. If you only have a few title slides, or such, to do, consider using a film recorder at your local lab. If you are going to want to make many slides, continue on to the details below. ___________________________________________________
THE DETAILS 1. Center your camera. It makes it easier if the monitor's flat screen is perpendicular to the table it sits on. Use of the test patterns (below) helps. Centering is important to minimize keystoning. A good working distance (via a mild telephoto) is useful so you can operate the keyboard and view the screen without bumping the camera. If the camera is farther away it is unlikely you will see reflections of it in the image. In all cases darken the room.
2. Include in your test roll some shots of a full screen pattern like this. When you take the test roll, note down the coverage of the viewfinder (e.g. which lines appear at the edges) and compare with the actual images in order to compensate for viewfinder coverage errors. Download a full size (2048x1556) version of this (big file). 3. Using the monitor controls, set its contrast to 60%, brightness to 60%, and color temperature to 5000K. Turning down the contrast is critical. Turning down the brightness insures that blacks will be black. Set your display driver to its full resolution (2048x1556 is best, but 1600x1200 or 1280x1024 can make decent slides too, just scale down the numbers on the resolution chart below to see what you might get). 4. Make a full screen image out of R = G = B = 100. Use your SLR's exposure meter setting as a basic exposure. Try for a shutter speed of 1 or 2 seconds (f stop typically will be 8 to 11 or so). Long shutter speeds are necessary to avoid seeing refresh artifacts. 5. Shoot a test roll of typical subjects along with a test-pattern image. Bracket exposure by + and – 1 stop (½ stop steps) to check exposure. You might want to try my baseline color correction. For Provia 100F I found the best overall color was obtained if I correct the screen image by: Color Balance (in photoshop or paintshop pro or equiv) -14B +7R, and apply an unsharp mask filter with settings 100%, 1.2 pixels, threshold 3. Depending on the subject a saturation increase of 10 or 20% might help. Do these latter things before you adjust the contrast and brightness of your monitor so you can see the results properly. After you have made the contrast and brightness and color temperature adjustments to your monitor all your pictures will look kind of dull and odd on the screen.
My Test Image. For a full resolution version (2048 x 1556) click here (big file). 6. For final color correction, use a method similar to making filter corrections in old wet-darkroom Cibachrome printing. Return to the monitor setting you use to enjoy your source images. Create a double window within Photoshop. Put the source image in one, and a Gray 200 in the other. Hold up the test roll slide image over the Gray 200 window (backlighting it). Compare the two. Adjust the color of the Gray window (using the color balance adjustment function of Photoshop or equivalent) until a decent match is obtained. Compensate the source images by the same percentage changes in its color balance. Example: You need -40 B on the Gray window to get a match. Looking at its histogram, you see that this amounts to a 20% decrease of B in the gray window (i.e. 40 / 200). Then make a –20% adjustment of B in the source image relative to the peak (or mean) of its B histogram. Example: If the mean B of the source image is 100, use –10B on it. Same for R or G if 2 colors need correction. For non-uniform monitors see the Appendix below. RESULTS The following results were obtained with a flat-screen 2048x1556 color monitor, a Pentax with an 85mm lens at f8, a 2 sec exposure, and using Provia 100F (normal development). For comparison purposes 35mm slides from the test roll(s) were digitized at full resolution (4000dpi) using a high quality film scanner. This may introduce some color shift but is the only way I could think of to indicate the nature of what you can get using this method.
The slide I got of the test image, re-digitized using a film scanner. Colors are pretty good. There's a little barrel distortion you probably could get rid of using the monitors controls - on a scene you wouldn't notice it. My 85mm lens is a cheap zoom. Registration and distortion errors will be the same provided you shoot everything you need (e.g. stereo pairs) in one sitting.
Digitized version of a nature slide. This was my "original" that was used on my test roll as a "typical" subject. This is a pretty tough image: high detail, high contrast, wide gamut.
Re-digitized image obtained from the "shoot you monitor" test roll using the settings suggested above. Color and contrast are pretty good. Further tweaking could improve things a little, but this was good enough for me.
Yuk. Here's what you get if you don't turn the contrast & brightness on your monitor way down, nor correct the colors. Unacceptable!!
Here is a blow up of the original (left) and monitor shoot (right) showing about 250 pixels out of 5500 (across). Obviously detail is lost between the digitized file (5.5K pixels wide) and the slide from a 2048 wide monitor. But it is not too bad. See resolution chart below.
On the left is a blow up of the slide output from an MGI Sapphire 4K professional film recorder (done at Photocraft Laboratories in Boulder, CO). On the right is the monitor-shoot result. I can't see much difference!
Here is a blow up of a photographic test pattern. This was obtained by shooting the pattern raw on Provia 100 then digitizing it. The upper right pattern on the inner block (1 - 1) is resolved. With this degree of magnification you can see some of the film grain.
Here is the monitor shoot result. I didn't worry about color in these tests. You can see the monitor pixels on the film. Block 0 - 1 (lower right of inside group), is resolved to about the same degree as the block picked out of the original. INTERESTING: When this slide is projected using an Ektagraphic and a Kodak (plastic) zoom projection lens you can't see the pixels, even up close! (With high quality projection lenses you can).
The MGI Sapphire 4K film recorder isn't much better than the shoot your monitor gig. This shows that a 4K addressable film recorder (which should ostensibly be equivalent to a 4096 pixel monitor) actually only delivers about half its quoted CRT addressable resolution. An 8K device does a little better BUT NOT X2!
Another way to get digital images onto film is to submit high resolution files to a LightJet 2000 (or 2080, etc) laser film recorder. This writes images on large sheets of transparency film by scanning RGB laser beams across the film at 2000 continuous tone dpi. Many slides must be laid out on an 7" x 9" canvas, then when you get the 8" x 10" film sheet back you have to cut them out and mount them by hand. Above is a blow up (about 250 pixels out of 5500 wide) of a test image.
Here is the LJ2000 result. Probably the best of the lot. Considerable degradation, but somewhat better than the CRT film recorder or monitor shooting. Our results are summarized below:
a with 4000 dpi dedicated film scanner.*MGI Sapphire 4K ** MGI Solitaire 8K (lower cost is if you lay them all out in powerpoint and get a quantity discount) *** Effective number of horizontal pixels (approx) Of course, the comparison with the Sanyo UXGA is a little unfair, because it can light up the 1600 pixels at almost 100% of its full contrast, with huge brightness. The monitor shoot is essentially almost as effective as the dedicated CRT film recorder, and falls a little behind the more expensive and complicated-to-orchestrate Lightjet scenario. References: Dan's Tips DPReview Forum Appendix: If you are unlucky enough to have a monitor with strong variations in illumination, here is one way to make a correction. Non-uniformity is usually not a problem except for source images with large monochrome patches (see my test result, above, which, if you look closely, is a little brighter on the right side). If you want to compensate for such, try this:
Repeat these steps for the Blue and Green channels in b) g) Save the modified image, which should be brighter where the monitor is darker, etc. This is a real hassle. If you stick to complex scenes (e.g. not like the viewfinder grid) you'll probably be OK without messing with this. |
