Stacking exposures to reduce noise?

So I met this guy at a friend's party who like to do astrophotography (sp). He was telling me that during a night, he'd take over 1000 shots and stack them on top of each other so the noise is reduce and he'd get a very clear photo. Has anyone every tried this before?

I stacked two 20-minute exposures once to lessen the noise and get some very long star trails. I never tried doing it as hundreds or thousands of separate photos, though. As little noise as I get in 20 and 30-minute exposures, I wouldn't think that's necessary.

This seems to be very popular in astrophotography and very effective. The 20D even has a form of it if you enable long exposure mode or whatever they call it.

Noise is random. By using multiple exposures of the same exact scene one can then compare each shot to each other and filter out the static stuff (the desired picture) from the random (the noise).

2 shots is certainly enough to seriously improve the quality, but 1000 simply seems crazy - though this would certainly enhance the barely-visible but static parts, like very faint stars. I'd go for something in between - but it totally depends on what I'm shooting. If I were to have a good tracking telescope then more is better, but in a general night scene 3 to 5 should be sufficient I'd think.

Taking that many "shots" is usually done with a webcam, and each video frame is viewed as a separate shot. There are several methods for adapting webcams to work with a telescope (the TouCam Pro has about the best reputation for quality). There are a couple of good software packages that automagically process the frames, and with several hundred to several thousand frames you can pretty well eliminate everything that isn't "picture". Using this method makes it possible to get some phenomenal results without absolutely perfect polar alignment and expensive tracking equipment.

Hmmm...interesting.
Well I have a rebel xt and I think i'm limited to how long my exposures are (or am I?) . If I wanted to try this could I do it with photoshop or would I have to get some specialized software?

Hope I'm using the right terminology here amd don't confuse myself or others.

What about dark field noise reduction? Where one leaves the lense cap on and takes an exposure equal to the longest exposure time you plan to use durring the real shoot, and subtracting this from the image. So, in essence you would end up taking out noise that is due to the perfromance of the sensor.

I've always heard it called Black Frame Subtraction, but it probably has many names. I've done that also when I did a few 20-minute exposures with my 10D. My 20D has so little noise that even on 30-minute exposures it was easier to just clone out the half dozen or so hot pixels in Photoshop than waste another half hour making a black frame exposure or using the 20D's built-in noise reduction, which does exactly the same thing, just in camera instead of on the computer.

I was very surprised at how clean the 30-minute exposure from the 20D was. One of these days I want to take it to a full hour and see how it performs. I can't even take my Olympus C-8080WZ to three minutes and still get a usable image!

UncleDoug wrote:
Hope I'm using the right terminology here amd don't confuse myself or others.

What about dark field noise reduction? Where one leaves the lense cap on and takes an exposure equal to the longest exposure time you plan to use durring the real shoot, and subtracting this from the image. So, in essence you would end up taking out noise that is due to the perfromance of the sensor.

Unless you have long exposure noise reduction turned off on your camera, this may be nothing more than an exercise in futility as the camera already does this automatically for long duration exposures. Since I'm one to try try anything at least once, even futile ones, I tried this already a while back and didn't get any meaningful results. It sounded good on paper when I first read about it, though. I think it was Scottes who brought up the subject.

Yep, I tested out the 20D's noise reduction and found it to be less than desirable for the amount of time it took. I can't find that thread though.

There's a big difference between Canon's long exposure noise reduction and stacking multiple frames though. Canon's way takes another exposure of the same length as the first, but with the shutter closed. 1 exposure of a lens cap produces random noise once.

Again, once.

One sample of randomness is insufficient, and does not leave any room for error or comparison. Stacking will produce random noise many many times, and clean picture many many times. The success ratio is VERY high. Flip a coin 4 times and you will often get results that do not produce Heads 50% of the time. Flip that coin 200 times and you will undoubtedly be very close to 50% Heads. So I never really bought into Canon's exposure of a lens cap theory too much.