A question on binning

[dodgerroger]

Hi there folks, I have just taken a big leap from dslr to mono cmos,A hyper cam 183.

i had the filter and camera all set up on my desk I thought I’d try a quick image plan in APT just to see how it all works. Anyway I set my RGB filters to 2x2 bin as people seem to use that for colour and 1x1 for luminance. Set it all going and APT came up with “ 2x2 binning not supported changing to 1x1?? 
I am I missing something??

[ollypenrice]

I wouldn't over-estimate the number of people who bin colour. I never do and I'm certainly not alone. Since you're using a CMOS camera I gather the advantage from binning is not what it is with CCD but I've yet to use a CMOS first hand. What is your pixel scale? There's a calculator here: http://www.12dstring.me.uk/fovcalc.php

I can't help on the software problem, not using APT.

Olly

[vlaiv]

16 minutes ago, dodgerroger said:

Hi there folks, I have just taken a big leap from dslr to mono cmos,A hyper cam 183.

i had the filter and camera all set up on my desk I thought I’d try a quick image plan in APT just to see how it all works. Anyway I set my RGB filters to 2x2 bin as people seem to use that for colour and 1x1 for luminance. Set it all going and APT came up with “ 2x2 binning not supported changing to 1x1?? 
I am I missing something??

Just shoot at 1x1 setting in your capture application.

Cmos sensors bin in software, and if you need to bin depending on your sampling rate - I can tell you how to do it.

9 minutes ago, ollypenrice said:

I gather the advantage from binning is not what it is with CCD

That sort of depends. Only difference between CCD hardware and any sensor software binning is read noise. You keep single read noise with CCD while you get double read noise for software bin (2x2 bin - x2 read noise, 3x3 bin - x3 read noise, .... etc).

In fact, depending on how you do software binning - it has other advantages, but let's not get into that now.

Only thing read noise is relevant to - is exposure length, and only in very dark skies without LP and when comparing high read noise camera, you can't "neutralize" read noise by choosing suitable exposure length (as in such conditions - you would need hour long exposures to "neutralize" read noise).

Plan on binning in software? Increase sub duration slightly (so that "diluted" sky noise still beats read noise on single pixel prior to binning and you are fine).

[dodgerroger]

50 minutes ago, ollypenrice said:

I wouldn't over-estimate the number of people who bin colour. I never do and I'm certainly not alone. Since you're using a CMOS camera I gather the advantage from binning is not what it is with CCD but I've yet to use a CMOS first hand. What is your pixel scale? There's a calculator here: http://www.12dstring.me.uk/fovcalc.php

I can't help on the software problem, not using APT.

Olly

My pixel scale is 1.18 I think

[dodgerroger]

So out of interest, if I wanted to bin colour it wouldn’t be in APT but at what point would it be done? I’ll keep it at 1x1 for now while I get my head around it all

[vlaiv]

46 minutes ago, dodgerroger said:

So out of interest, if I wanted to bin colour it wouldn’t be in APT but at what point would it be done? I’ll keep it at 1x1 for now while I get my head around it all

1.18"/px is on the edge to be over sampling. You need really good seeing / guiding / larger aperture to fully exploit high sampling rate.

What software are you using? If you are using PixInsight for example - calibrate your subs normally, and then prior to stacking - use Integer resample tool with average method (that will bin). Reduce subs that you want by x2 (for example color, although you might do both luminance and color).

Afterwards - proceed normally as if you captured particular set of subs with hardware binning.

[dodgerroger]

7 minutes ago, vlaiv said:

1.18"/px is on the edge to be over sampling. You need really good seeing / guiding / larger aperture to fully exploit high sampling rate.

What software are you using? If you are using PixInsight for example - calibrate your subs normally, and then prior to stacking - use Integer resample tool with average method (that will bin). Reduce subs that you want by x2 (for example color, although you might do both luminance and color).

Afterwards - proceed normally as if you captured particular set of subs with hardware binning.

I am using pixinsight at the minute, something else to get my head around. I thought the 183 was a good match for a short focal length scope

[vlaiv]

53 minutes ago, dodgerroger said:

I am using pixinsight at the minute, something else to get my head around. I thought the 183 was a good match for a short focal length scope

It is, but there are short focal length scopes an short focal length scopes.

It matches well with focal lengths of 320-350mm. Those are 70-80mm scopes in F/4-F/5 flavors. As soon as you get over 400mm you are starting to sample at close to 1"/px. That is about as good as amateur setups are able to do with high end mounts and larger scopes (8"+).

~ 80mm -> sampling rate of about 2"/px

~ 120-150mm -> sampling rate of about 1.5"/px

~ 8"+ -> 1 - 1.2"/px

would sort of rule of the thumb in sampling rates. You also need a mount that is capable of half your sampling rate in RMS or less. So for 2"/px you need something like 1" RMS mount (Heq5 does that fairly easily). 1.5"/px - you need 0.75" RMS or less - EQ6/HEQ5/CEM60 all do it if properly tuned and with some experience with guiding. 1-1.2"/px - you need 0.5 RMS or less. We now hit premium mount territory here.

All is not lost, with the scope you have and that sensor, you can easily bin x2 to get 2.4"/px for very good sampling rate.

In fact - you can check ideal sampling rate for your image - take any stacked image (or a single sub) and measure average FWHM in that image  - convert value to arc seconds (if in pixels, multiply with 1.18 to get arc seconds) and then divide value with 1.6  - that is optimal sampling rate for that particular image.

Let's say you have 3.2" FWMH stars - optimal sampling rate wil be 3.2" / 1.6 = 2"/px

One of advantages of software binning is the fact that you can choose to bin if it makes sense - by examining your image after you capture it. If night was particularly good and seeing stable and your mount well behaved - small FWHM - leave image unbinned if it makes sense - otherwise bin it.

[dodgerroger]

Yeah that does make sense, good job I have pixinsight then. Thanks for the info. So if I can summarise in my laymen terms shoot 1x1 in APT and then follow your earlier explanation in pixinsight to get 2x2👍. All good fun eh thanks again. 

[vlaiv]

4 minutes ago, dodgerroger said:

Yeah that does make sense, good job I have pixinsight then. Thanks for the info. So if I can summarise in my laymen terms shoot 1x1 in APT and then follow your earlier explanation in pixinsight to get 2x2👍. All good fun eh thanks again. 

Indeed. It costs you nothing to try. If you select bin 2x2 in APT - it is irreversible action - you can't split binned subs to unbinned. However, you can shoot at 1x1 and then if you choose to do so - bin to x2 - it won't be any different compared to capturing it originally as binned x2.

Only difference when binning in software is that you have choice and you can compare what you prefer - make stack with original 1x1 or bin each sub after calibration and stack that and process both. Some people don't like to bin and find it easier to process over sampled image because they never look at image in full zoom. Having larger pixel count lets them apply more sharpening and image looks better / sharper on screen size (but it looks much worse with fat stars when you zoom in to 1:1 to see the details).

[dodgerroger]

Great advice. This is a bit of learning curve compared to my old canon 600D

[michael8554]

I seem to remember reading that Lum is shot at optimum resolution, but RGB does not need to be high res so can be shot binned, which reduces exposure times.

Michael