Page 6 of 8

Re: Picking the correct exposure for Deep Sky

Posted: Mon Mar 07, 2022 2:48 pm
by Borodog
Hmm. How does that work mathematically?

Take for example my ASI183MC. Shooting in might light pollution (B5) at 4/f.8, this difference in optimal exposure at 0 gain and unity gain is about a factor of 3. The difference in DR is about 1.5 stops, about a factor of 2.7. The FWC at 0 gain is 15k; at unity it's just over 4k. So 3 unity gain exposures don't get me quite back to the same capacity.

Not trying to argue; just trying to understand.

SharpCap's Smart Histogram has 2 selectable gain targets, MDR & unity. But many people say that unity is not really a special or magical gain. Why not half unity? Or twice?

A separate consideration for my camera is that it is subject to banding that is worse at lower gain, so that alone might be a reason to choose unity over zero. Although I haven't tested SharpCap's banding suppression for the 2 different gains.

Re: Picking the correct exposure for Deep Sky

Posted: Mon Mar 07, 2022 5:20 pm
by admin
Hi,

if there is a sharp drop in read noise as the gain is increased then the minimum exposure length needed can drop *a lot* - halve the read noise and the exposure length drops by a factor of 4, so you can take 4x as many frames in the same length of time. Each of them may have a bit less dynamic range than the low gain single frames, but the fact that you have 4x as many can overcome that.

This definitely happens for cameras that have a switchover from LCG (low conversion gain) to HCG (high conversion gain) at a particular gain value, since the step decrease in read noise is usually enough to cause this to happen. For cameras without that step drop in read noise, it all depends on the relative rates of change of the dynamic range and read noise as the gain changes.

chreers,

Robin

Re: Picking the correct exposure for Deep Sky

Posted: Mon Jun 20, 2022 6:18 am
by barnold84
Hi,

I'm having some confusion here about the read noise. As stated in the beginning, we can safely assume that the read noise follows a Poisson distribution. Say that this distribution has an expectation value of lambda and therefore, the variance is lambda (and standard deviation sqrt(lambda), which is often abbreviated with sigma).

Now, when the vendor states the read noise, I would assume that this would be our lambda? If so, going back to the equations in the first posts, what am I using for sigma_r or sigma_r^2 respectively to compute the exposure time?
My understandng would be that the read noise sigma_r^2 = lambda = stated read noise of camera?

e.g. read noise = 5e-, sky electron rate = 1.5e-/s, allowed noise 5%, hence tmin = 1/(1.05^2-1) * 5/1.5 = 32.5s?
Or is it tmin = 1/(1.05^2-1) * 5^2/1.5 = 162.6s?

Björn

Re: Picking the correct exposure for Deep Sky

Posted: Mon Jun 20, 2022 1:28 pm
by admin
Hi Björn

you have your noises mixed up I think... It's the shot noise that follows the poisson distribution, with the standard deviation being equal to the square root of the number of electrons measured. This noise is not a property of the camera, it's simply down to the quantization of photons and the electrons they are converted into.

The read noise of the camera is (probably) normally distributed and is a property of the camera, independent of the number of electrons being captured at an individual pixel.

Hope this helps,

Robin

Re: Picking the correct exposure for Deep Sky

Posted: Tue Jun 21, 2022 4:05 pm
by barnold84
Hi Robin,

I guess the question if something is Poisson is not the most important aspect here. The justification would anyways be done if the underlying process is about rare events. In the limit, the Poisson distribution converges into a normal distribution.

My question is indeed what to understand about the term "read noise" or to be precise the underlying process causing this noise. My model was always to see read noise in the way that the reading event is "creating" additional electrons, i.e. read noise which would be modelled by a Poisson process. So from what you're indicating is that the causing process is a different one and if we would always read a pixel with a constant number of electrons (since we prepared a pixel in such a way), we would create a gaussian variation where the standard deviation is the read noise and independent of the actual number of electrons in the pixel?

Björn

Re: Picking the correct exposure for Deep Sky

Posted: Wed Jun 22, 2022 12:54 pm
by admin
Hi Bjorn,

yes, that's how I understand it - the read noise is a combination of the fluctuations in the analogue electronics and the A->D conversion that eventually turn the pixel brightness from a number of trapped electrons into an ADU number. This is independent of pixel brightness.

The read noise is also uncorrelated with the poisson shot noise, which means that the two noise sources need to be added in quadrature (square each, add, take square root). This allows the read noise to have practically zero impact once the shot noise level exceeds it by a factor a factor of 5 or so.

cheers,

Robin

Re: Picking the correct exposure for Deep Sky

Posted: Wed Jun 22, 2022 1:03 pm
by barnold84
Hi Robin,

Thanks for clarification. I'll try getting a bit more of physical background of read noise.

I don't want to spill this topic with other issues, so please let me know if I should move this to a different thread:
your making the point of swamping the read noise to which I fully agree. I'm still wondering a bit why some people work with second or sub-second exposures for DSOs when the read noise is hightly significant. They never would reach the SNR that long exposures do, or am I missing something? I can't imagine that it's all about detail (w.r.t. to resolution) and a lot of equipment would at least allow 30 or 60 second exposures (guided and some unguided).

Best,
Björn

Re: Picking the correct exposure for Deep Sky

Posted: Wed Jun 22, 2022 2:06 pm
by admin
Hi Bjorn,

I think you are right that the use of very short exposures will never get good SNR (possibly except on very bright targets with large pixels and fast optics). On the other hand, you can pick up exceptionally sharp images if you can freeze the seeing variations to some extent.

cheers,

Robin

Re: Picking the correct exposure for Deep Sky

Posted: Thu Sep 01, 2022 4:41 am
by Ana
I am in bortle 7. Smart histogram shows 10 sec exposure for Celestron 8; asi294mc, optolong L-pro filter. This is good enough to get picture for globular cluster (M5 or M13), but this set up absolutely not enough for M80 cluster or M81/M82 galaxy. I don't see details of the objects.
Let's take the same set up and run: 1 min 180 subframes or 10 sec 1080 subframes. First, my computer will struggle to integrate and stack 1080 frames. But, if it manages, which picture (1 min 180 subframes or 10 sec 1080 subframes) will have better overall outcome (more details, sharper details, high SNR, etc).
My experience show that 1min 180 subframes will be better. Why?

Thanks

Re: Picking the correct exposure for Deep Sky

Posted: Thu Sep 01, 2022 12:44 pm
by oopfan
Hi Ana,

Have you tried stacking 1 min exposures? Is it really that much better than 10 sec?

The Smart Histogram is telling you the minimum exposure. You should not go below it, but you can always go above it for a small improvement. If you go too high however then you can potentially run into problems with clouds, aircraft, poor tracking, star saturation, etc.

Can you post your M81/M82? Perhaps there is something besides exposure that is the culprit. There are many factors that make a great photo. My M81 required a stack size spanning 11 hours. Generally, I find that many astrophotographers underestimate how long it takes.

Brian