Can you expose for separate RGB channels with the 40D?

Just a curiosity question... Looked through the manual, and was unable to find an answer. Was thinking about using it for astronomy applications if it could be done.

no.

hmmn, maybe use a colored filter to block some wavelengths?

Just curious what your purpose would be in doing this.

You can change the histogram to display RGB levels, I use that to check white balance since I'm partially colour blind.

Ehhh... Not really sure why my prof was asking, but I think it is to see different emissions from different wavelengths. The red channel is the most sensitive, but even the green and blue channels can pick up details that the red may not.

Actually, your image is a 3 by X by Y matrix. X and Y are your image dimensions and 3 corresponds to RGB. Each element of the matrix is a quantized intensity level by 13 bits or 14bits. So if you want to isolate each channel, you can do this with Photoshop.

You can separate the channels in photoshop, but you can't do this in the camera (all three will be captured with each exposure).

Although, saying that, it's possible to have your camera modified so that only a particular channel is recorded on the sensor (I believe this is how IR camera conversions are done)

But Pete, that would mean I'd have to have at least 3 different 40D's!!!

Actually. What you need to do in order to fulfil your request is to look at the different Red Green and Blue channels in your post processing software of choice.

Uh, maybe I'm missing something here. Can't you just use a red, green or blue filter on the lens to get this result?

If you want a full-color image exposed for the red channel, set exposure with the filter and then remove it for the shot.

Or do I misunderstand?

-js

You don't want to use a color filter because it unnecessarily blocks out other light. Just separate the channels in Photoshops.

Can you expose for separate RGB channels?
If by expose, you mean determine exposure, then yes, through use of the RGB histograms.
If you mean record only one channel, then no, but it doesn't matter as you can just throw away the two you don't want.

Would filters help? Not really (but see below). There are already filters over the sensor.

In short, you can produce images containing any blend of the signals passing through the three different bandpass filters over the sensor. What their cut-off characteristics are I don't know.

What you can't do, either in the camera or in post production, is implement any other band-pass filter. You cannot, for example, do spectral analysis. You cannot isolate particular emission bands. For that, you would need additional notch filters in the optical path.

I have hit this problem in fluorescence microscopy. It is impossible to distinguish something that has yellow fluorescence from something that fluoresces both red and green. This can be overcome by recording an emission spectrum for each pixel, but the amount of data is vast, the processing time high, and the expense great. This facility is often incorporated into laser scanning confocal systems though, from Leica, among others.

So - you can have R, G, or B as your camera now stands. With filters, you can have any narrow band you want recorded within the encompassing R G or B channel. With modification, this can be extended to IR and UV.
What you can't have is spectral analysis on the cheap.

I don't know, what "expose for separate channels" means.

Re separating the channels before demosaicing: it is possible, though tiresome.

Separating the channels in Photoshop etc. is a totally different issue: those R,G and B components are very different from the raw channels.

One should keep in mind, that the raw channels are not R, G and B but Rgb, Grb and Brg. All three kind of filters over the sensels transmit most of the visible spectrum, but to different degree. Even monochromatic light, like laser or LED is recorded by all three channels; the relative intensities of the recorded pixel values yield the original color.

Added filters can not separate the light waves perfectly either. Beside, what is "red light" or "green light" or "blue light"? There is no single red, etc. - these are ranges of light wave lengths with washy boundaries (and what about the yellow light?).

What I am curious about is using the camera more like a CCD for a telescope. Guess I should have been more clear. I'm not really sure what my prof was interested in, other than he wants the blue and red channels of the Pleiades. And no nebulosity, just the stars.

Your prof sounds like he wants to use some narrow-pass color filters to isolate out the nebulosity of the Pleiades (I work in a space research center).

Xarqi gave a good answer and my opinion is that you really can't do what your prof's looking for, at least not very well. Your DSLR's image sensor has built-in non-removable filters over the individual pixels and any filters you try to add will only be in series with these. For a better explanation of this, look up "Bayer Pattern" and you'll see how this works in almost all digital cameras:

http://en.wikipedia.or​g/wiki/Bayer_pattern

Consumer digital cameras are made to record three fairly broad-band colors. I don't know of any consumer cameras sold without these built-in filters but you can buy fairly inexpensive CCD or CMOS imagers made especially for astronomy that will do just what your prof wants, maybe for a few hundred to a couple of thousand dollars. They have no filters built-in and require a filter set or filter wheel, which gives you great flexibility in color imaging and low-res imaging spectroscopy. I think he's trying to do this on the ultra-cheap!