OK - Here's my initial thoughts on the subject. (I love instrument and analysis design, by the way.)
I see two basic approaches. One can either:
1) make something, or
2) adapt existing equipement.
With that in mind, let's look at Option 1. Going with Patrick's excellent suggestion of using silicon phototransistor would be the first step. Some method would be needed to convert the voltage signal from the op amp into something usable, i.e. a volt meter at the least.
1.a) Adding on to the phototransistor idea, some sort of an optic path would be needed to isolate and focus on a small section of a negative. And a light source as well. And finally, some sort of mechanism like a micrometer to move the negative.
For this approach, probably basing the design off of traditional microdensitometers would be best. That means a microscope-like type of optics would be needed.
This may be a lot of work. Data collection from the volt meter and then conversion to density readings could be painful for the number of readings that would be needed.
Perhaps easier, would be:
1.b) Keep the negative in a fixed location, and move the sensor. Some sort of mechanism could be used to move the sensor under a projected image of the neg. Using an enlarger (which most of us already have) to focus and project the image onto the baseboard, a baseboard densitometer like the phototransistor/volt meter combo or perhaps easier, something like on of the old Minolta PM densitometers could be used. The PM (or other baseboard meter) would be direct reading in density, so that would simplify the conversion of voltage readings in to density.
Things to keep in mind with this approach:
1.b.1) Move the enlarger up as high as it goes to increase the image size, as well as using a shrt focal length lens to increase the amount of magnification at the baseboard. (We don't need the lens to cover the entire neg, just something that will project a sharp and contrasty portion of the negative.)
1.b.2) Mask the opening of the sensor to get a smaller aperature to further increase the effective magnification. Ideally, the size of the aperature should be somewhere around the step size of the linear distance that the sensor will be moved.
1.b.3) Use a set threaded rods (i.e. long screws) to make a "sled" for the sensor to sit on, and then a "dial" could be added so that fine turns of the rod could be measured. A 1 mm pitch rod being turned 15 degrees per step would be 1/24th of a mm increments. Multiple by the baseboard magnification to get the actual step sizes on the negative.
1.b.4) Use of the color filters of the Minolta PM (or other) baseboard densitometer unit could be used to isolate color/stain effects.
For Approach 2 - using existing equipment, off hand, I can think of:
2.a) find a microdensitometer of some commercial equipment that is similar in function. If actual microdensitometers were easy to come by and cheap, we would not be having this thread...
But I think Tom has a good approach - use a microscope with a digital cameral and take images of the film. I spent about 2 hours trying this one afternoon at a friend's house who is a microscopist. He had a nice Zeiss scope and a digital camera that is designed for use with the scope. I had issues with getting enough magnification while still keeping the image sharp. The contrast was not what I was hoping for as well. And then I had problems getting the image bright enough to shoot with the camera. (I burned a few holes in the test neg a could of times before I figured out how to not do that!)
But I think this idea has a lot of merit. But it is not a cheap solution for those of us that do not have access to expensive scopes and digital cameras. So on to the next idea:
2.b) Film Scanners - I think this apporach should be a good one, and since many of us already have scanners, then that is one big hurdle down as far as the equipment side of this problem goes.
2.b.1) Resolution - I have a Nikon Coolscan V that can do 4000 dpi. That's about 157 points per mm. Looking in James and Mees' microdensity plots, that looks like it should be sufficient, if not just what is needed. Even though that scanner only does 35 mm format, cutting negs up to feed it should not be a issue.
2.b.2) Data Collection - another big plus for the scanner, simply scan the film. Once the scan is in the computer, then we can use software to get the density readings.
Take a look at this page:
http://www.efg2.com/Lab/ImageProcessing/TestTargets/ (this will be useful for our next big project on how to test for resolution!) and then notice the graphs at the bottom of that page - they used the software that can be found here:
http://www.efg2.com/Lab/ImageProcessing/PixelProfile.htm
I've played with the Pixel profile software and it is pretty simple and easy. Bring a picture into it, drawn a line across the image, and then it makes a spreadsheet page full of data from the pixels under the line.
The issues I can forsee with this are not enough resolution in the scanner's density readings - scanners are designed for macrodensity and probably not as much for microdensity. I have a feeling that 8-bit resolution will not be enough. So perhaps 12 or 14 or 16 bit image files would be usable. (I don't know if the Pixel-Profile software can handle these files, then then we can do stuff in photoshop to extract a set of pixel data I'm sure.
And in general, Ron and Patrick are right that calibration issues will be important. Especially after reading Dr. Henry's account of his attemps at microdensitometery in "Controls in Black and White Photography".
And targets as well. Perhaps Ron could elaborate on the target he has. I was thinking that a USAF 1951 contact target for resolution would go a long ways here, but the chrome on glass ones that would be best are pretty spendy. I figured photos of the large USAF 1951 target that Edmunds Scientific sells would be good for starting - it has high and low contrast patches, as well as R, G, and B ones.
Hope this gets the ball moving along on this idea! See, I told you I like this sort of stuff!
Kirk -
www.keyesphoto.com