What is the point of combining images in astrophotography?
bastian915 asked:
I've seen photos where people have combined say 10 15 minute exposures of a nebula and I can't figure out why….are they somehow thinking that by adding one layer to another they are somehow increasing the amount of exposure?? I can understand using different filters and sandwiching them together to build up a colour image but exposure I just can't wrap my head around it. Another example was this picture of the gabriela nebula and the photographer said he exposed the HA layer for 19 hours but there just isn't 19 hours of dark. I'm thinking someday it would be nice to get into astrophotography but I just can't wrap my head around the long exposures and combining images (other than just combining multiple filtered images together that's easy).
"Stacking exposures is indeed the same as taking one long exposure " that is what I can't seem to understand because if you have two photos with the same luminesence I can't see putting the two together having a duplicating effect
I've seen photos where people have combined say 10 15 minute exposures of a nebula and I can't figure out why….are they somehow thinking that by adding one layer to another they are somehow increasing the amount of exposure?? I can understand using different filters and sandwiching them together to build up a colour image but exposure I just can't wrap my head around it. Another example was this picture of the gabriela nebula and the photographer said he exposed the HA layer for 19 hours but there just isn't 19 hours of dark. I'm thinking someday it would be nice to get into astrophotography but I just can't wrap my head around the long exposures and combining images (other than just combining multiple filtered images together that's easy).
"Stacking exposures is indeed the same as taking one long exposure " that is what I can't seem to understand because if you have two photos with the same luminesence I can't see putting the two together having a duplicating effect
June 12th, 2008 at 9:08 pm
Combining images reduces noise, or graininess, in the final image. It basically smooths the rough image and allows more sharpening to be done, making a clearer image.
June 15th, 2008 at 6:58 pm
Any photographic medium is going to be subject to error. Any given pixel of a CCD , or particle of photosensitive material in a piece of film might be slightly different on any given exposure to the same light.
Using several images allows for greater accuracy - and thus simulated higher resolving power - by averaging the results. You could even create a higher-resolution composite than your originals were. The differences in pixels can be extraoplated to determine fairly accurate values of "in-between" regions, too.
The more frames you use for your composite, the more accurate it can become. Of course you want the frames to be as similar as possible to begin with. If one is overexposed, it will throw the average off, for instance.
June 17th, 2008 at 9:10 am
Yes, by adding up the images, they are increasing the exposure. CCDs are very linear, and one 10 minute exposure will capture just about as much light as ten 1 minute exposures added up. And be a lot easier to manage, because the short exposures sharply reduce guiding requirements.
Second, there is always random noise in CCD images. When you add the frames, the noise is random and cancels out. The image is not random and adds up.
Addendum: each exposure captures so many photons, resulting in so many electrons in each well of the CCD. It doesn't matter if you do the exposure all at once, or do several shorter exposures: you will end up reading off the same net number of electrons, corresponding to how much light the CCD gathered.
June 17th, 2008 at 12:21 pm
You are really just counting photons. The more photons you get per pixel, the better the signal to noise ratio of your image and the cleaner it looks. For a given instrument aperture (and a given focal length) the only way to increase the number of photons is to wait longer and let them accumulate. And since it is easier to make a large number of short exposures than a single long exposure, astronomers "stack" images. The statistics does not care how it's being done. But the short term stability of instruments is much better than their long term stability, so stacking and suitable correction algorithms for images give the most bang (signal quality) for the buck.
June 17th, 2008 at 3:48 pm
Yes, combining many images does increase the amount of exposure, and will increase both the sensitivity and the dynamic range of the image.
June 20th, 2008 at 9:58 pm
Stacking exposures is indeed the same as taking one long exposure. You just keep adding the brightness values together. Taking numerous short exposures enables you to throw out ones with problems like passing airplanes, guiding errors, or whatever, and can reduce the guiding precision required. It also avoids overexposing bright areas. Most of all, it enables you to do noise reduction. The noise will be different in each frame, while the signal remains the same.
A 19 hour exposure is obviously accumulated over a period of several nights.
The links below will give you some idea of what's involved.
Added: The addition of exposures works like this. If you have a 12-bit readout on your CCD, each pixel can have a luminance value of from 0 to 4095. A short exposure of a dim star might give you only a value of 10, which is essentially black. Adding 50 of these exposures adds those values up, so you end up with a very visible 500. And while there might be equally bright bits of noise on each exposure, they don't add up, because the noise is random. Even if you just added up the noise with the signal, it would average out to a slight haze. But you can digitally filter out all the bright spots that only occur once to further clean it up. On the other hand, just taking the short exposure and multiplying all its values by 50 will equally amplify both noise and signal.
You can do similar stacking with film, where each exposure will add a little more density at the location of the star.
June 22nd, 2008 at 3:20 am
I'll re-iterate here the fact that it's for increasing SNR (Signal to Noise ratio). The best possible case is to take one single long exposure. However, this is in many cases unfeasible. Sometimes hours of exposure are desired. Taking a single exposure for a couple hours has many problems. First it would probably overexpose the sensor, making it worthless. Second, your equipment would have to be very steady. Any movement or shift would ruin the whole exposure. It also increases the chance of things like an aircraft or satellite passing through the field.
Each exposure you take (with a digital camera anyway) has some read noise in it. Read noise depends on the camera and not the image itself, so it's got a constant amplitude regardless of the length of the exposure. A single long exposure captures read noise only once. Multiple exposures capture it multiple times. So, a single hour long image will have a somewhat higher SNR (assuming it didn't overexpose) than 6 stacked 10 minute exposures. So while injanier is mostly correct, he is technically wrong on the point that stacking multiple exposures is exactly the same as taking one long one. If that was true, then we'd all just take thousands of 1 second exposures and stack them.
June 23rd, 2008 at 7:51 pm
Your question (and confusion) seems to be why taking many shorter images when you can simply take a long exposure and be done with it. If so, here is my answer.
There are at least two major factors affect SNR:
Atmospheric emission and its variation (on scale of minutes)
Sensitivity variation across the CCD detectors
A single long image would suffer both of them, especially when imaging the faint objects. Imagining a cirrus flow though the field of view for a few minutes during the 5 hour exposure, that would ruin the whole picture. Or you detected many more stars on the top of the image just because that's where the sensitivity is higher.
Mindful of these problems, astronomers usually take many short exposures while dithering the telescope show each image is offset for a small distance. In post-processing, sky background emission is subtracted out from, and CCD QE calibration was done on each image. The images then are realigned and stacked together. That's when the excitements usually come when you start to see faint features in the combined image, that's not present in the individual ones.