1100d ISO or exposure length

[Solitarian]

Just curious as to others views for imaging in which should be given priority, ISO or exposure length and how many images are worth stacking? Can you have too many to stack ?

Tested out a Vixen Polarie Sunday night with an 1100d attached, started at 60secs and worked up to 12 minutes with ISOs from 800 to 3200. Ended up with a mixed bag but I'm not up with processing so I may not be seeing what others see in an image.

The Polarie was what I was focusing on and it's tracking capabilities, I just dropped the ISO down as the exposure lengths increased or if the images were too bright.

With the 1100d being a fairly common camera here, before I give it another try I thought I'd ask others advice on what direction I should take in regards to the exposures. What I should looking for

There's a couple of images from Sunday night attached, just adjusted levels, all the same way, nothing else.
2 and 4minutes at 1600 and 5 and 12 minutes at 800

Cheers

[LightningNZ]

Unfortunately it's hard to tell what difference, if any, there is in such small images... So here's the rub, for images that you're going to scale down massively anyway (e.g. 12 megapixel to 2.5 megapixel 1920x1200) you can bump up the ISO a long way because the noise you get will be smoothed out when you average several pixels in the original. The amplification provided by a higher ISO however doesn't get you more signal, it just moves the signal you have towards the brighter (right-hand) end of the brightness histogram, making what you have easier to see.

Ultimately there's no substitute for lots of exposure, but perhaps as can be seen by the last image in your line-up, sky glow will eventually drown out long images, so there is a dizzy limit. Then you just stop exposing and take another picture. Stacking these will smooth out the noise and let you pull more signal from your images.

I hope that was reasonably clear. I don't think there are any hard and fast rules to this.

[raymo]

You can stack as many as you like. Most people seem to stack around
25-100, but some people stack as many as 2 or 3 hundred.
raymo

[Solitarian]

Thanks Cam, that's very clear, I just need to work within the limits of each, I'll do a few more test sessions next week when the moon rises later and see whats best.
Cheers

[dannat]

is tht a lot of light pollution in the bottom of frames? if thats the case you might want to stack a few shorter mages or get a lpr filterif i remmebr 1600 was usable on the 1100d but 3200 pushed it too much..check the noise levels on some 100% crops of your images

[Solitarian]

Thanks raymo. I wasn't sure if there was a point of diminished returns, so on a good night it would be worth putting the time in to gather extra images, that's good to know.

Cheers

[Solitarian]

It's actually the neighbours floodlights bouncing off my shed, they have them on every night till about ten. I can work around them when I need to, they'll even turn them off if I ask, nice neighbours actually. I wasn't concerned as I was more interested in testing the Polarie that night, which worked well.
3200 is noisy, I even thought 1600 was noisy compared to 800, but it's easily disguisable .

Cheers

[jsmoraes]

You must pay attention to histogram. Place the curve distant from left with 50% of the width of the histogram curve.

And we can say (it is not absolutely equal) that 10 min with ISO 400 = 5 min with ISO 800 = 2.5 min with ISO 1600 = 1.25 min with ISO 3200.
The problem with high ISO is the noise.

Stack image: the more the merrier, better result with singal/noise ratio. But there is a useful limit, e. g. 1,000 frames will not be much better than only 100 frames.

Some programs of stacking can cause problem with the colors of stars if you use many frames: all them will be white, or with better words: saturated. DSS is one of them.

[LightningNZ]

Because in each frame you have a fixed noise component (mostly the sensor read noise in the camera) the extra signal-to-noise you gain increases with the square root of the number of subs you take. So if you take 4 you double your SNR (signal to noise ratio), but taking 25 only gives you 5 times the SNR, 100 subs-> 10x...

So taking more subs is always good, but in terms of just increasing the SNR more exposure in each sub is more effective - until you max out the signal in your sensor (stars are bright and max out quickly) or the background brightness of the sky is the only thing you're continuing to enhance - the so-called "sky fog".

[Merlin66]

I know we have discussed this at some great length previously, but I still feel the major "issue" with AP images is the "unwanted signal" rather than just the SNR calculation....
( Coming from spectroscopic observations, where the final SNR in the recorded spectrum is critical. The aim is to achieve at least a SNR =>40.
We can't (for scientific reasons) manipulate, correct or play around with the with spectral images other than the standard flats/ darks and sky background (pollution removal))

[Solitarian]

Are you saying to aim for the curve to be positioned from the left, half the width of the curve?

[Solitarian]

Gotcha, eventually the effort will exceed the rewards, a case of learning when enough is enough.
Thanks again.

[Solitarian]

What is the unwanted signal you're referring to comprise of, sorry, beginner here.

[LightningNZ]

I presume Ken means all the nasty non-linear things we do to an image to make it look "pretty". Deconvolution, sharpening and weird histogram manipulations - things you couldn't even do in a darkroom. If that's the case I agree, less can be more. Either way I prefer to see final images that "respect the light" and are not too different in interpretation from what is "actually there".

[Amaranthus]

Unwanted signal is mostly sky glow (background light pollution), lunar wash out, zodiacal light etc. Everything that makes the sky brighter that is not the target DSO!

[jsmoraes]

Yes. If the histogram curve is together the left axis, the image will be very dark. If the histogram curve is very distant, the image will be very bright.

The curve of histogram shows the intensity and amount of signal that you got.

To take it in a good balance you must consider the width of that curve. If you take the half of that curve as the ideal distance to the left axis, try ISO and time of exposition to get it.

The image will not be dark neither bright. It will have a good balance: a natural appearance. To enhance details or area, you can do it with graphic processing.

[Solitarian]

Thanks Jorge,
I'll play around and work with that on a few images, it's all new to me.

Cheers