Trichromy

Trichromy is a process of producing a color photo by taking three separate monochrome (black and white) photos then combing them into one color photo later. This process is much like early color photography which required three exposures; one through each of the primary additive colors: Red, Green, and Blue. This means that you end up taking three separate exposures taken using a red, green, and blue filter respectively.

You have to use a sturdy tripod to take three photos of a scene in which there is little to no movement (assuming you don't want any color artifacts)

You can combine these three separate images into one color image using Photoshop or another advanced image editor.

How to combine the channels into one composite image in Photoshop
In order to combine them into one color image, open them all in PS. Name each one appropriately as Red, Green, or Blue. Then take one of the images, select the Red channel, right click it, select Duplicate Channel, Create New, name it Red. Go to Image > Mode and change the mode to Grayscale for each of them. Save each of the three files to the same directory as TIFFs (I don't think that you have to do this, but it is a good idea should something go wrong). Then, go to one of the grayscale files, let's say Red. Make sure the other files are also open so you should have three files open now, one named Red, one named Green, and one named Blue. Select one of the images. Go to the Channels Palette. Hit the little hidden menu icon in the corner to revel the drop down menu. Select Merge Channels. Then select RGB Mode. Make sure it is set to 3 colors. Then hit OK. It should automatically list the three channels that will make up the color version. Make sure all 3 channels are indicated in the correct order. This will then give you full color image. Areas that didn't move should look normal and those parts that moved will have different colors to them. You may have to align all three of the layers to account for any that were recorded out of register.
NOTE: You will get a slightly different look by changing the images from which you took the respective primary color from. That is to say, you can take the Red, Green, and Blue channels from a different one of the original color images. This will result in the colors being different.
Have fun with this!

Still trying to solve problems with white areas

In an effort to insure the problematic white areas aren't caused by stray light from the built in LED light sources, I removed the little panel which contains all the lights from the sensor array. Much to my relief, the scanning back still functioned using VueScan.
Prior to this, I installed Ubuntu Linux on a partion of my hard drive. I used the open source SANE driver to drive the scanner in hopes that the white areas wouldn't be there. They were still there.
I ordered a set of LEE 3" primary color filters a few days ago. Kit includes a Red, Green, and Blue filter that are very restrictive so that they only allow in their respective set of wavelengths. This will allow me to create a full color image by taking three seperate shots through each of the color filters and then blending the images digitally on the computer. Of course, time will have passed between each shot, so many artifacts will occur. I am very excited about how this will look. Furthermore, since the sensor is sensitive to infrared light, that will affect things as well. Look forward to seeing what I get.

Quick overview of the project

Welcome to my blog! I am writing this in an effort to dispense information regarding the creation and utilization of digital cameras made from modified scanners.

A little background. I am currently a MFA Digital Arts student at Bowling Green State University at Bowling Green, Ohio. For the past eight years I have taught digital photography at various schools and colleges. Having used 4x5 scanning backs in the past and recognizing their similarity to scanners, I wondered if it would be possible to convert a flat bed scanner to function as a scanning back. The short answer is yes. After looking into it, it turns out that there are others that beat me to the punch.

My efforts have been largely been based on the work of others.

Michael Golembeski has a site dedicated to his scanning camera work that has been helpful. http://golembewski.awardspace.com/index.html

A Phd student in computer science also did some excellent work that I found very helpful. Here is a link to his thesis about a color scanning camera. http://www.cs.ubc.ca/~heidrich/Papers/EG.04.pdf

I will add links to other relevant sites in the near future.

By the way, if you don't know what a scanning back is, you can visit the Betterlight website which is one a few companies still producing scanning backs (they were much more popular at the advent of professional digital photography) http://www.betterlight.com/ They have some excellent information and resources there.

The traditional camera portion of my scanning camera outfits consists of a 1950's Deardorff 8x10 field camera in need of restoration and an assortment of vintage lenses. My moderate wide angle lens is a 1950's Kodak 190mm f6.3 lens. I also have a Goerz Artar 16.5" f9.5 process lens. This lens was intended for use with a horizontal process camera for reproduction use so it doesn't have a shutter. It is from the 1930s or 1940s. Lastly, I just got a Schneider Super Angulon 165mm f8 back from repair. It is from 1957 or 1958. It is a beautiful super wide angle lens.

The scanning back is made from a modified Canon Canoscan LiDE 25 flatbed scanner held in place by a simple wooden support. The scanner is an inexpensive consumer model that retails for around $50. This scanner lends itself particularly well to being modified for use as a scanning back as the linear sensor array is the actual width of the scan meaning that is roughly 8" tall. Most scanners make use of a sensor array that is significantly smaller than the width of the scanner. They make use of mirror and/or lenses to scale the image down to the array's size.

Essentially, the modification to the scanner consists of removing a little strip of pinholes, which served to funnel light to the sensor, disabling the light source, and widening the area over the sensor to allow light in from oblique angles to be recorded. Once the scanner is modified in this fashion, the Canon scanning software will no longer drive the scanner. This is because the Canon software performs numerous self checks that the modified scanner fails. So, at the moment I am using VueScan to drive the scanner. However, I plan on switching to using Linux and using SANE to drive the scanner instead. Not there yet.

Anyways, I have included a few samples of what I have come up with so far. The first image is a crop from a larger image. I am getting white areas on either side of the scan. The second image shows a scan with the lens cap on. The third image is a full frame scan. The white areas are in the same exact spot as the previous image. Notice the black bar area along the right hand side. Since my camera is an 8x10 camera, it has an area of roughly that size onto which light strikes. The scanner scans at 8x14 though. So, the black area is an area where there is no light striking the scanner.

The white areas are not light leaks. The scanner can be rotated 180 and they rotate 180 as well. They have something to do with the calibration of the scanner. All scanners with which I am familar go through an initial calibration routine. First, they perform a dark calibration in which the scanner determines black by taking some samples with no light. Then, it performs a white calibration in which it determines white by taking some samples of the white lid with the light on. I have not been able to perform a successful calibration routine with the scanner camera setup. In order to obtain usable calibration information, I have to use an unmodified scanner to generate it then switch over to the modified scanner. So, the modified scanner is using the calibration data from an unmodified scanner. Not elegant, but the best I have come up with so far. I am looking for a way to ignore the calibration routine altogether.

I am firmly commited to this project. I plan on using this camera and the work I am creating with it for my MFA show. I will be modifying this post and adding others in the near future. Thanks for visiting!