#1  
Old 10-10-2015, 05:34 PM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
To bin or not to bin

I've never binned RGB before (actually never done much RGB at all until recently), but I'm questioning whether I should start doing it.

Analysis of recent images using PI's Sky limited exposure script suggests the following unbinned exposure times:

L = 480s
R = 956s
G = 1461s
B = 2992s

So that's about 38min exposures for blue to be at the beginning of "sky limited". I've been aiming for an arbitrary 5 subs RGB lately, with 30+ for L. Doing a mere 5 blue subs is over 3hrs, and that's assuming none of them get trashed. While I'm sure I'd have great data if I did it, I'm not sure if I have the patience for that just yet.

I'm considering binning my RGB, and then drizzling it 4x (L will be drizzled 2x) to combat this, but I'm not sure if this would be doing me more harm than good.

I had been adding my RGB to the L to improve SNR, but obviously if I start doing this I'll no longer be able to do that.

Thoughts?
Reply With Quote
  #2  
Old 10-10-2015, 09:25 PM
gregbradley's Avatar
gregbradley
Registered User

gregbradley is offline
 
Join Date: Feb 2006
Location: Sydney
Posts: 17,877
Binning is useful at times.

1. When you want less noisy colour and time is limited.

2. Seeing is not so great.

3. You use it for Ha or O111 for increased colour in your neb/galaxy image.

4. The image is not sharply detailed so its not as important to try to get maximum detail.

5. It increases full well capacity so star sizes can actually be tighter in some instances. ICX674 38 minutes would surely be oversaturated unless its dim Ha.

I use 1x1 when I want maximum detail even in colour and I have the time to get enough to smooth out the extra noise it has.

2x2 on say a 4.54 micron pixel makes it a 9 micron pixel and super high QE and full well which would work well on a long focal length scope.

I think this covers all the uses I am aware of.

Some CCDs do not do a proper 2x2 binning. The KAF8300 is only capable of 50,000 electrons when binned when it should be capable of 100,000 so its 2x2 binning is half of other cameras that do a true 2x2 and not limited in the electronics like the Sony ICX CCDs and I think most Kodak sensors (the KAF8300 may be the odd sensor out there).

Greg.
Reply With Quote
  #3  
Old 10-10-2015, 11:34 PM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
 
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
you can test it out on your existing data and see if it works well enough for you. http://pixinsight.com/forum/index.php?topic=3195.0

Then, if you think it is OK, try it in hardware. You will probably find that low res (binned) colour is OK without drizzle - interested to hear what you think.

I think you will find that the 674 does not give you the theoretical 80,000 electrons/pixel in 2x2 bin mode, because 80ke cannot fit down the transfer registers or through the output amp. expect maybe ~40ke (possibly less) before saturation. However, the read noise should be close to normal when binned, so you still get the sensitivity advantage, even if not the full dynamic range advantage.

Last edited by Shiraz; 10-10-2015 at 11:45 PM.
Reply With Quote
  #4  
Old 11-10-2015, 08:18 AM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
Thanks guys, appreciate the replies. I guess the basic question is: does getting above the read noise improve the colour data more than the loss of resolution detracts? Hard to say I guess.

I'll have a look at doing a software bin soon and see what it looks like. Won't be quite the same; I'll have a bit more dynamic range from deeper wells and less read noise, but it should be in the ballpark.

The other question is: how important is it really to be sky limited. I know it's ideal, but when we're talking dark skies with a camera that has only 5.4e- of read noise, makes me wonder whether it's really worth it...
Reply With Quote
  #5  
Old 11-10-2015, 08:24 AM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
 
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,918
if your subs are not sky noise limited, you can get the same result using more overall exposure time to compensate for the extra read noise. the sole driver behind optimising sub exposure is to minimise the time needed to get to a given image quality. you may wish to choose other exposure times - eg to improve dynamic range with shorter subs or because you don't want to risk something screwing up and losing lots of data in long subs..

Last edited by Shiraz; 11-10-2015 at 09:27 AM.
Reply With Quote
  #6  
Old 11-10-2015, 08:36 AM
Slawomir's Avatar
Slawomir (Suavi)
Registered User

Slawomir is offline
 
Join Date: Sep 2014
Location: North Queensland
Posts: 3,240
Hi Lee,

To my knowledge, with RGB filters the usual goal is to represent "natural" colours of the stars and DSOs. So I thought you would want to have similar exposures for each filter corrected of course for CCD's QE for each colour. ICX 674 has higher QE for green and blue colours than in the red, so perhaps exposures through red filter need to be a tad longer. It also depends on filters transmittance of light as well.

I would recommend finding a right RGB exposure ratios for your RGB filters and perhaps expose for as long as possible with the Lum filter.

EDIT: I think it makes sense to me to avoid saturating stars with RGB filters
Reply With Quote
  #7  
Old 11-10-2015, 08:46 AM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
Cheers S.

Yeah, there's something unusual going on and I'm not sure what it is. If you look at the spectral sensitivity curve for the 674, you're right, green is the most sensitive, although not by a huge margin.

And if you look at the bandpass for the Astronomik filters, they're almost equal.

However, when calculating a sky limited exposure you get (obviously going to depend on the method you use), the numbers I posted above. Oh and it's not a bug / error in that calculation either; the background ADU of the subs back up those numbers.

So either the sensor doesn't follow the curve it's supposed to, the filters don't have the bandpass they're supposed to, or the (dark) sky here is a lot more red than I'd have expected.

End of the day, doesn't matter which one of those it is if I don't change anything.

The ratios can be easily calculated from the above, scaling as appropriate.
Reply With Quote
  #8  
Old 11-10-2015, 09:48 AM
alpal's Avatar
alpal
Registered User

alpal is offline
 
Join Date: Jan 2012
Location: Melbourne
Posts: 3,602
Quote:
Originally Posted by gregbradley View Post
Binning is useful at times.

1. When you want less noisy colour and time is limited.

2. Seeing is not so great.

3. You use it for Ha or O111 for increased colour in your neb/galaxy image.

4. The image is not sharply detailed so its not as important to try to get maximum detail.

5. It increases full well capacity so star sizes can actually be tighter in some instances. ICX674 38 minutes would surely be oversaturated unless its dim Ha.

I use 1x1 when I want maximum detail even in colour and I have the time to get enough to smooth out the extra noise it has.

2x2 on say a 4.54 micron pixel makes it a 9 micron pixel and super high QE and full well which would work well on a long focal length scope.

I think this covers all the uses I am aware of.

Some CCDs do not do a proper 2x2 binning. The KAF8300 is only capable of 50,000 electrons when binned when it should be capable of 100,000 so its 2x2 binning is half of other cameras that do a true 2x2 and not limited in the electronics like the Sony ICX CCDs and I think most Kodak sensors (the KAF8300 may be the odd sensor out there).

Greg.

Don't tell me that - I use a KAF8300.
Reply With Quote
  #9  
Old 11-10-2015, 09:51 AM
Slawomir's Avatar
Slawomir (Suavi)
Registered User

Slawomir is offline
 
Join Date: Sep 2014
Location: North Queensland
Posts: 3,240
Hi Lee,

As you have already hinted, with green filter you will have significantly higher background signal, followed by blue, and red should have the lowest background signal for the same exposures. It makes sense to me that with brighter background you need to expose for longer, while with perfectly black background you will be only overcoming noise from bias, thus exposures can be shorter

EDIT:
(From http://www.stanmooreastro.com/eXtreme.htm )

(...) Thus bright objects are not much harmed by light polluted skies.

(...)for dim objects, the time and sky terms nearly operate in tandem. For example, a 2x brighter sky requires almost 2x more exp time to reach a similar dim object S/N. A typical semi-urban/suburban sky will have a brightness of approximately 17 mag/arcsec^2 and a good rural sky is about mag 20. That means that a dim object will require approximately 16x more exposure time under the bright sky to equal the S/N of the dark sky.

Last edited by Slawomir; 11-10-2015 at 10:16 AM.
Reply With Quote
  #10  
Old 11-10-2015, 10:49 AM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
I think I found the issue. SX reports a different spectral sensitivity curve than does Sony for the same sensor. Sony's graph is more in line with what I expect.

The peak of the curve is in the gap between Astronomik's red and green filters, which means the red will actually end up more brightly exposed than the green. The blue is also different on the graph, but in both it'd be the darkest and I think the numbers roughly correlate to the background values in the subs I'm seeing.

As for the brighter skies need longer exposures... that's very interesting and goes against what a lot of people say.

As I understand it, the reason to aim for being "sky limited" is because shot noise and target signal both increase with exposure time, but your read noise is effectively static (random, but not increasing with exposure time).

Being fully sky limited means that your background signal swamps the read noise... so at this point if you're sky limited at 1min, 10x1min = 1x10min in terms of SNR on the background.

This is where it gets interesting, and your quotes from Stan seem to be the opposite to what most people recommend?

My question from earlier was why try to be "fully" sky limited in dark skies with a low read noise camera... I'm going to blow out a ****load of stars trying to do that.

Shouldn't I really be aiming to be sky limited on the minimum target value, rather than the background? Maybe add a bit of buffer for ease of processing.
Reply With Quote
  #11  
Old 11-10-2015, 12:22 PM
Slawomir's Avatar
Slawomir (Suavi)
Registered User

Slawomir is offline
 
Join Date: Sep 2014
Location: North Queensland
Posts: 3,240
Hi Lee,

I think CCD's spectral sensitivity varies too little to really explain why in your case the blue filter needs 300% of the exposure of the red filter (from your original post) to be sky-limited.

What you are saying about sky-limited theory makes sense to me, but it does not explain the numbers we are getting from various methods. In my case, I am supposed to expose for MUCH longer with O3 filter (weaker signal) than with Ha filter (stronger signal).

In case of narrowband imaging exposures and signals from various filters do not need to be balanced as nicely as from RGB filters, where we aim for natural colours, that's why I suggested not worrying about sky-limited exposures for RGB, but rather focus on getting a correct exposure ratios for those filters and not saturating the stars if the goal is accurate colouration of the stars.

I think the sky-limited theory does not take into account well depth. So the length of exposure ultimately depends on your objectives: long exposures for maximum SNR or shorter ones for non-saturated stars.

Since with RGB filters we are after colour, shorter exposures make sense to me.

Last edited by Slawomir; 11-10-2015 at 12:37 PM.
Reply With Quote
  #12  
Old 11-10-2015, 12:36 PM
RickS's Avatar
RickS (Rick)
PI cult recruiter

RickS is offline
 
Join Date: Apr 2010
Location: Brisbane
Posts: 10,584
Quote:
Originally Posted by alpal View Post
Don't tell me that - I use a KAF8300.
Greg didn't give you the other bad news... you don't get much of an improvement in read noise when binning on the KAF-8300 (or KAF-16803 and probably other Kodak/TrueSense sensors.) They work perfectly well unbinned which is how I always use them...
Reply With Quote
  #13  
Old 11-10-2015, 12:38 PM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
Hmm, no, you're right. I'd expect to see about 30% longer exposures for B, not 255%.

I've now realised that the Astronomik LRGB filters I bought are type 2, not 2c. I can't find any data on the type 2, so I'm not sure what their bandpass might be. Strange though, I wouldn't expect it to be that much different.
Reply With Quote
  #14  
Old 11-10-2015, 12:40 PM
alpal's Avatar
alpal
Registered User

alpal is offline
 
Join Date: Jan 2012
Location: Melbourne
Posts: 3,602
Quote:
Originally Posted by RickS View Post
Greg didn't give you the other bad news... you don't get much of an improvement in read noise when binning on the KAF-8300 (or KAF-16803 and probably other Kodak/TrueSense sensors.) They work perfectly well unbinned which is how I always use them...

I use mine binned for Ha & RGB.
Maybe I should change & use 1x1?

I am still undecided about binning.
Binning at 2x2 should increase the signal to noise ratio according to studies I have read.

Also -
Think about it -
if you use narrow band such as Ha then you can pick detail out that is lost in a Luminance image.
Also you can image in light polluted skies with narrow band & still get top results.

Why not consider RGB as a form of narrow band?
Why take Luminance at all in light polluted skies?
If you just took RGB at binning 1x1 & made a pseudo luminance frame from it
then you would get better signal to noise ratio in light polluted skies.
I will try this & I have already noticed when I process that if I use 100% Luminance
in Photoshop then the image is worse.
I usually have to reduce the opacity of the Luminance frames.

cheers
Allan

Last edited by alpal; 11-10-2015 at 12:52 PM.
Reply With Quote
  #15  
Old 11-10-2015, 12:51 PM
RickS's Avatar
RickS (Rick)
PI cult recruiter

RickS is offline
 
Join Date: Apr 2010
Location: Brisbane
Posts: 10,584
Quote:
Originally Posted by codemonkey View Post
This is where it gets interesting, and your quotes from Stan seem to be the opposite to what most people recommend?
The quote from Stan above was talking about total exposure time, not individual subs so I don't see any inconsistency?

Quote:
Originally Posted by codemonkey View Post
Shouldn't I really be aiming to be sky limited on the minimum target value, rather than the background? Maybe add a bit of buffer for ease of processing.
If your minimum target value isn't anything slightly higher than the background then you're not trying hard enough
Reply With Quote
  #16  
Old 12-10-2015, 07:21 AM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
Quote:
Originally Posted by RickS View Post
The quote from Stan above was talking about total exposure time, not individual subs so I don't see any inconsistency?
Ah ok, given the context of the post I had assumed it was individual sub duration.

Quote:
Originally Posted by RickS View Post
If your minimum target value isn't anything slightly higher than the background then you're not trying hard enough
Good point... I had a look at the faintest part of NGC 300 that I could distinguish from the background and there was bugger all difference; it was closer than I thought.
Reply With Quote
  #17  
Old 13-10-2015, 08:15 PM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
I suspect I know what the discrepancy is here: it occurred to me that when imaging targets low on the eastern side, I image in order of R, then G, the B, so that the B subs, which I believe should be more affected by poor seeing, are shot through less atmosphere.

Since the images were taken hours apart, with R first, it's likely that the sky was simply a bit lighter then.
Reply With Quote
  #18  
Old 14-10-2015, 05:02 AM
Peter.M's Avatar
Peter.M
Registered User

Peter.M is offline
 
Join Date: Sep 2011
Location: Adelaide
Posts: 947
Quote:
Originally Posted by codemonkey View Post

I'm considering binning my RGB, and then drizzling it 4x (L will be drizzled 2x) to combat this, but I'm not sure if this would be doing me more harm than good.
Just so people are aware, the second you drizzle your data your SNR goes down. For me I use drizzle when I have LOTS of subs and SNR is less of an issue. But when your trying to maximize low amounts of data, I dont think it will be a good idea
Reply With Quote
  #19  
Old 14-10-2015, 06:19 AM
codemonkey's Avatar
codemonkey (Lee)
Lee "Wormsy" Borsboom

codemonkey is offline
 
Join Date: Jul 2013
Location: Kilcoy, QLD
Posts: 2,058
Thanks Peter, that's a really good point. I actually reprocessed my recent NGC 300 last night for that very reason and it made a big difference because of how few RGB subs I had.
Reply With Quote
  #20  
Old 15-10-2015, 11:03 AM
gregbradley's Avatar
gregbradley
Registered User

gregbradley is offline
 
Join Date: Feb 2006
Location: Sydney
Posts: 17,877
Quote:
Originally Posted by RickS View Post
Greg didn't give you the other bad news... you don't get much of an improvement in read noise when binning on the KAF-8300 (or KAF-16803 and probably other Kodak/TrueSense sensors.) They work perfectly well unbinned which is how I always use them...
My FLI Proline has a choice of download speeds, 1mhz and I think its 8mhz. The lower read noise figure for that camera is the 1mhz one. Do you use slower digitization rates on your Apogee and get the lower read noise?

I am wondering if I should shift over to doing that. 1 second downloads are seductive but a few extra seconds for lower read noise is not much to ask I suppose!

Greg.
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 06:35 AM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Advertisement
Bintel
Advertisement
Testar
Advertisement