Recommended a refractor please

[Pitch Black Skies]

Ah perfect. So am I correct in saying that Bin 2 will be faster than the unlocked Bin 1 mode? But Bin 2 and Bin 3 will actually be the same brightness as they are within the same mode? Hope that makes sense. 

[vlaiv]

49 minutes ago, Pitch Black Skies said:

I thought by using Bin 2 I was going to make the system 5x faster.

It is, but with cmos sensors - it does not matter if you bin it during capture or later in processing - result is the same.

I know this might be a bit counterintuitive - but if we put it in right context - it will make sense.

Binning is nothing more than stacking applied "horizontally".

With regular stacking - we average a number of captured subs - meaning pixels in each sub at certain coordinates - say 201, 136 - get averaged out between subs. Key here is average (or sum) of pixel values.

Same thing happens with binning, except pixels are not in different subs at same position - they are in same sub at different positions. We average / sum groups of pixels. This of course has drawback of doing something to resolution / total number of pixels - but if you are over sampling or generally don't care about finest detail - it will perfectly fine.

Stacking is done after capture, not during. Similarly - binning can be done after capture and it will produce same effect.

2 hours ago, Pitch Black Skies said:

Would it not be better for you to bin at data capture instead? It would reduce your imaging time.

You can reduce your imaging time if you plan to bin in software later.

Better yet - look at it this way, you image for the time you have at your disposal, and binning will just improve SNR that you are able to achieve in given time.

If you don't bin in software and leave it at bin1 - it will have corresponding SNR, but if you bin x2 - you will improve SNR by factor of x2. If you bin x3 - you will improve SNR by factor of 3.

[vlaiv]

24 minutes ago, Pitch Black Skies said:

Ah perfect. So am I correct in saying that Bin 2 will be faster than the unlocked Bin 1 mode? But Bin 2 and Bin 3 will actually be the same brightness as they are within the same mode? Hope that makes sense. 

Binned data can have same brightness as not binned data, or it can be brighter than non binned data - depends on how you do the binning.

If you use average - it will keep the same brightness. If you use sum - it will be "brighter".

It will however improve SNR in both case. Remember - brightness is just something you assign to numerical value. You can say my image has pixel value of 100 units, or you can say - my image has pixel value of 1unit per second, or you can say - my image has value of 60 units per minute.

Those are all same thing even if they have different numerical value assigned to it. Similarly - by adjusting white and black point - you say, this 60 units per minute will be brightest part of the image.

Since you can't separate noise and actual pixel values - if they are close and you make signal bright, you will make noise bright as well and it will be seen in the image. This is why signal to noise ratio is most important thing. It let's you set signal brightness without making noise bright enough to be seen.

Hope that makes sense.

[Pitch Black Skies]

Thanks, apologies but this one isn't sinking in fully yet.

I am under the impression that binning at data capture allows me to collect more photons (super pixel) AND improve SNR, but at the expense of finest detail. 

Where's if I'm to bin in processing, my assumption is that the photons have already been captured, so now I ONLY have the opportunity to improve SNR, but at the expense of finest detail.

[vlaiv]

44 minutes ago, Pitch Black Skies said:

Thanks, apologies but this one isn't sinking in fully yet.

I am under the impression that binning at data capture allows me to collect more photons (super pixel) AND improve SNR, but at the expense of finest detail. 

Where's if I'm to bin in processing, my assumption is that the photons have already been captured, so now I ONLY have the opportunity to improve SNR, but at the expense of finest detail.

You are almost there

Number of captured photons is very tightly related to achieved SNR.

Binning at capture time - does not make you capture more photons. You would need larger aperture for that. What it does (same as larger pixel) - is to rearrange how those captured photons are distributed - how many of them is "in each pixel".  However - that does not change if you sum them after measuring or before measuring.

Imagine you have four buckets containing golf balls.

In one case - you take your four buckets and pour all of them in one large bucket and then count golf balls (binning at capture time)

In other case - you count balls in each bucket and then add them up on paper.

Are you going to end up with different result? No - you'll count the same total number of golf balls. Way you count them is irrelevant.

Here is comes important bit. I mentioned that number of photons is related to SNR.

Larger number of photons - larger the SNR and there is mathematical relationship.

It is exactly the same if you do the following:

capture 100 photons in single pixel (large pixel)

capture 25 photons in each of 4 pixels and then "pour" them in single bucket and count them -again 100 photons - same SNR

capture 25 photons in each of 4 pixels and them mathematically add those numbers - again 100 photons and same SNR

expose for x4 longer with single small pixel - you capture 100 photons - again same SNR

capture 25 photons in each of 4 smaller pixels and mathematically average those - here you will have resulting number be 25 but that is because you only changed "unit" - you have 24 "4photons" - and same SNR. (SNR does not change if you divide both signal and noise with 4 when you average 4 samples - divide both numerator and denominator with same number - fraction remains the same).

That is why stacking works - it is like adding exposures together to get the same number of photons as would be captured with one single long exposure.

(above is strictly true if there are no other noise sources except shot noise, but SNR part remains the same - stacking won't be the same as one long exposure when read noise is non zero, but that is another topic).

[Pitch Black Skies]

Thanks, you have explained that very well. I'm getting there. I still feel a small bit stupid though 😅.

The light gathering formula is what's  confusing me. 

My bin 1 figure of 6,272 and bin 2 figure of 25,344.

I'm reading that as bin 2 collects ~4 times more light, thus lets me reduce my total imaging time by a factor of ~4. Albeit with half the resolution.

It has me thinking I could capture an image in bin2 for 1 hour,  and it would be the equivalent of a 4 hour image in bin1. But the bin2 image would be half the size. 

[The Lazy Astronomer]

This is very much vlaiv territory now, so I'm going to sit back and try to finally properly absorb the theory myself as well 😁

[vlaiv]

10 hours ago, Pitch Black Skies said:

I'm reading that as bin 2 collects ~4 times more light

I think that you probably should not read it like that.

What it should really say is either:

- collects x4 more photons per pixel for same exposure length (either sub or total - for both is true)

or

- needs x4 less time to achieve the same target SNR

It is formula for "speed" and not how much signal is accumulated over time. It just happens that "speed" is related to signal per pixel (not overall / total signal - that depends on aperture size alone).

Nothing really magical is happening here. It is all very rational if we use right words to explain what is going on.

If we bin 2x2 - we will get x4 larger pixel - so yes, it is completely rational that x4 larger pixel by surface collects x4 more light if aperture stays the same and total number of photons that hits the sensor does not change. Larger pixel will simply scoop larger part of those photons.

 

 

[Pitch Black Skies]

Perfect, I wasn't using the right words and that was confusing me.

I have noticed that when we bin x2, we are actually getting a fraction more than 4 times the photons per pixel. Does that mean that the QE slightly improves with bin x2 also?

So is there little point to binning in the field, other than reducing file size? Is it a good approach to bin x1 at data capture in the hopes of excellent seeing and if it's oversampled just to bin it in processing?

Edited September 7, 2022 by Pitch Black Skies

[vlaiv]

5 hours ago, Pitch Black Skies said:

I have noticed that when we bin x2, we are actually getting a fraction more than 4 times the photons per pixel. Does that mean that the QE slightly improves with bin x2 also?

Not sure what you've noticed, but no - binning just means summing up photons. You can't suddenly get more photons.

Doing it at capture time or doing it afterwards - result is the same. Binned pixel photons  = sum of individual pixels that are binned.

QE does not improve.

5 hours ago, Pitch Black Skies said:

So is there little point to binning in the field, other than reducing file size? Is it a good approach to bin x1 at data capture in the hopes of excellent seeing and if it's oversampled just to bin it in processing?

I prefer to capture at native resolution and then use any binning afterwards, and yes - I can decide on bin factor based on achieved resolution. For my gear it is usually x3, rarely x2 and sometimes x4

Another benefit is that you can choose how you want to bin your data. You can bin it normally, but you don't have to. Fact that we are stacking multiple subs gives us opportunity to do some fancy stuff.

We can do split bin instead of regular bin. That just means that we split pixels into separate subs instead of adding them up (they will be added / averaged in stacking anyway). This way we create 4 subs for each sub taken - and we end up with x4 more subs (as if we imaged for x4 longer - again it is all the same stuff really )  - this helps with something called pixel blur - but that is very advanced to explain and not very noticeable in everyday use, so you don't have to worry about it.

In any case - yes, only advantage to capture time binning (for cmos sensors) is smaller file size and faster downloads (but cmos download subs in less than a second anyway).

[Pitch Black Skies]

1 hour ago, vlaiv said:

Not sure what you've noticed, but no - binning just means summing up photons. You can't suddenly get more photons.

Just that bin 1 with my setup is 6,272, and bin 2 is 25,344.

When I divide 25,344 by 4 I get 6,336, so a little more than 6,272. I was thinking that extra 64 was an improved ability to collect more photons.

[Pitch Black Skies]

On 22/08/2022 at 17:20, scotty38 said:

Agreed on all of that although I've not yet used that mode on my 294m but I was only suggesting it based on your desire of sampling though.

I might have been wrong on that. It's probably not as bad as I was making out. The unlocked bin1 mode looks similar to bin1 on both the 1600MM and 183MM.

[vlaiv]

3 hours ago, Pitch Black Skies said:

Just that bin 1 with my setup is 6,272, and bin 2 is 25,344.

Can you run the math by me on that?

What numbers did you use for bin 1 and bin 2?

[Pitch Black Skies]

Bin 1 0.99"/pixel

Bin 2 1.99"/pixel

 

pixel^2 x aperture^2

= 0.99^2 × 80^2

= 0.9801 x 6,400

= 6,272.64 bin1

 

pixel^2 x aperture^2

= 1.99^2 x 80^2

= 3.9601 x 6,400

= 25,344.64 bin2

 

25,344.64/4 (dividing bin2 by 4)

= 6336.16

 

 6336.16 - 6,272.64

= 63.53

[Pitch Black Skies]

I've figured it out. The bin 2 should be 1.98"/pixel.

Not sure why the tool is saying 1.99.

Maybe just a glitch.

[The Lazy Astronomer]

15 minutes ago, Pitch Black Skies said:

Bin 1 0.99"/pixel

Bin 2 1.99"/pixel

 

pixel^2 x aperture^2

= 0.99^2 × 80^2

= 0.9801 x 6,400

= 6,272.64 bin1

 

pixel^2 x aperture^2

= 1.99^2 x 80^2

= 3.9601 x 6,400

= 25,344.64 bin2

 

25,344.64/4 (dividing bin2 by 4)

= 6336.16

 

 6336.16 - 6,272.64

= 63.53

I think the discrepancy is just a result of rounding errors

(80^2*(4.63/480*206.265)^2)/4 = 6333.599

80^2*((4.63/2)/480*206.265)^2 = 6333.599

Edited September 7, 2022 by The Lazy Astronomer
Made my own rounding error!