stacking images

[robatman (Robert)]

Just trying to get my head around why stack multiple images.

Assuming you have perfect tracking and same exposure per image theoretically each image made should be exactly the same. Why not just stack the same image?

I have also read a little about thermal noise and it would seem the last image can tend to be noisier than the first so the latter images will be of a lesser quality.

I get the need for darks (to map noise reduction) and i think flats (to aid in alignment and reduce distortion?)

What am i not getting?

Robert

[andrewk_82 (Andrew)]

Hi Robert,

Put simply, stacking images reduces noise by averaging it out. The more images in the stack the greater the level of noise reduction. I'm sure someone who knows a lot more than I do about it will expand on this.

A good way I was helped to understand it is. Imagine you draw a 5 x 5 grid (representing pixels) on some clear plastic and colour in the squares but randomly leave say 8 clear (this represents noise). Now do it again but leave a different lot of 8 squares clear. Do this say 5 times and then stack these grids on top of each other. Due to the random spacing of the clear squares the stacked image would be almost completely (if not completely) coloured with almost no noise (clear squares). The noise was averaged out due to its random spacing.

I hope this helps

Cheers

[Zuts]

Generally you are trying to increase the s/n (signal to noise) ratio. The noise comes from many sources, ccd read noise, thermal noise, cosmic ray hits and so on and is random in nature; while the signal, nebulae, stars and so on is constant.

Stacking amplifies the signal but randomises and reduces the noise, if you stacked the same image over and over you would also be amplifying the noise.

You wont see things that are not there, i.e. a 60 minute sub will always be deeper than 6 by 10 minute stacked subs but the signal visible in a stack of 6 by 10 minute subs will always be stronger than 1 ten minute sub; and the image will look smoother and less noisy.

Cheers
Paul

[multiweb (Marc)]

Originally Posted by robatman

Just trying to get my head around why stack multiple images.

Assuming you have perfect tracking and same exposure per image theoretically each image made should be exactly the same. Why not just stack the same image?

I have also read a little about thermal noise and it would seem the last image can tend to be noisier than the first so the latter images will be of a lesser quality.

I get the need for darks (to map noise reduction) and i think flats (to aid in alignment and reduce distortion?)

What am i not getting?

Robert

Hi Robert, loads of questions in there . In a nut shell stacking will increase your SNR (signal to noise ratio) and also allow you to reject bad pixels.

Darks are for minimising thermal noise (random). Flats are for getting rid of uneven illumination (vignetting) and all sort of dust motes, stains on your optics, etc... that stay in the exact same spots within your imaging run provided the camera and the focus haven't moved.

Stacking pictures minimise sky noise and other nuisance like satellite trails, asteroids, bad seeing. This is achieved in many softwares by running rejection algorithm so you keep the good data and reject the bad data. It is also achieved by slightly dithering (offsetting) your picture so when you realign them the "bad pixels" don't fall on top of each others and therefore become rejected over the number of sub frames you stack.

[leon]

Hmm Robert, that certainly is an interesting thought, so what you are saying is copy any amount of the original shot, and then stack them as if one had taken a certain amount of images of that object.

I like the thought

Leon

[Dennis]

Originally Posted by leon

Hmm Robert, that certainly is an interesting thought, so what you are saying is copy any amount of the original shot, and then stack them as if one had taken a certain amount of images of that object.

I like the thought

Leon

Hi Leon

I believe that stacking several copies of the same image will not improve the Signal to Noise ratio, or minimize the “grainy” effect in the photo, because the “noise” or “grain” will always appear in the same place on each copy and thus add up.

But, I do stack the same photo as several Layers in Photoshop and this can “improve” a photo. For example, one Layer may have the colour saturation boosted, another Layer may have some Noise Reduction applied, another Layer may have the shadows lightened and yet another Layer may have been sharpened.

When you the carefully blend these Layers together, you can get the benefit of the saturation, noise reduction, shadow detail and sharpening appearing in the final image and still control each layer to adjust how it contributes to the final image. But you can’t really improve the S/N Ratio with a single image.

Cheers

Dennis

[RobF (Rob)]

One other tidbit - the reduction in noise for a series of identical exposures is proportional to the number of shots (total exposure time). As you double the number of shots ("subs"), you can theoretically halve the noise in the final stacked image. However as you increase to longer and longer total exposure times, the incremental benefit keeps reducing.

All this depends on light pollution, noise characteristics on the camera, imaging temp and other nasty issue too in practice. But hey, if it was easy I think most of us wouldn't have got hooked on astrophotography.

[robatman (Robert)]

Thanks guys,
It makes sense now I think that noise occurs randomly and multiple shots can help the software or Post processing discriminate and then reduce noise.

Robert