#1  
Old 09-11-2018, 09:03 AM
Andy01's Avatar
Andy01 (Andy)
My God it's full of stars

Andy01 is offline
 
Join Date: Mar 2012
Location: Melbourne
Posts: 3,253
Use Bias, Dark Flats or both?

Hi folks,

I'm using a QSI683 wsg8 and recently had a go at using Astropixel Processor. Very cool software indeed.
APP allows the user to load lights, darks, flats, dark flats, bias or masters of each of these in any combination.

Some users are recommending using bias instead of dark flats - which I understand is very convenient.
Now I'm not looking for a very technical answer, I just want to use what works best! Usually I take lights, darks, flats & dark flats, but no bias.

So my question is - should I be using Bias or Dark flats? What's best?
Bias or dark flats or both?

cheers
Andy
Reply With Quote
  #2  
Old 09-11-2018, 09:40 AM
Atmos's Avatar
Atmos (Colin)
Ultimate Noob

Atmos is offline
 
Join Date: Aug 2011
Location: Melbourne
Posts: 6,980
I’d just use dark flats personally, even more so with a CMOS sensor; which the KAF-8300 isn’t.
Reply With Quote
  #3  
Old 09-11-2018, 09:51 AM
Merlin66's Avatar
Merlin66 (Ken)
Registered User

Merlin66 is offline
 
Join Date: Oct 2005
Location: Junortoun Vic
Posts: 8,904
+1 for the dark flats - as per AstroArt processing.
Reply With Quote
  #4  
Old 09-11-2018, 01:52 PM
RickS's Avatar
RickS (Rick)
PI cult recruiter

RickS is offline
 
Join Date: Apr 2010
Location: Brisbane
Posts: 10,584
I already answered you on Astrobin
Reply With Quote
  #5  
Old 09-11-2018, 03:01 PM
ChrisV's Avatar
ChrisV (Chris)
Registered User

ChrisV is offline
 
Join Date: Aug 2015
Location: Sydney
Posts: 1,737
Can someone explain what the issue is with using bias frames rather than dark flats. Is it because the shorter bias frames are quite variable compared to longer dark flats - and supposedly worse with CCDs? So introduce noise even if you average a fair few? Or is it something else ....

I use 0.2sec subs for my bias frames. Might that be considered to be a dark flat given my flat subs are about the same
Reply With Quote
  #6  
Old 09-11-2018, 03:12 PM
Merlin66's Avatar
Merlin66 (Ken)
Registered User

Merlin66 is offline
 
Join Date: Oct 2005
Location: Junortoun Vic
Posts: 8,904
Chris,
It's not necessary to deal with bias frames at all.

My understanding is that the longer exposure used to obtain a "good" flat requires its own dark to remove any/ all dark "thermal noise" related issues.

AstroArt Manual, p167
""But which role has the bias frames in our CCD Image processing? Each frame captured with a CCD camera contains the bias frame. Then, when we subtract the dark frame, we are subtracting at the same time the bias frame and we do not need to deal with it any more""

(The bias frame only comes into play when attempting to scale the dark from one exposure and/or temperature to another)
Reply With Quote
  #7  
Old 09-11-2018, 03:43 PM
Atmos's Avatar
Atmos (Colin)
Ultimate Noob

Atmos is offline
 
Join Date: Aug 2011
Location: Melbourne
Posts: 6,980
Typically CCDs only have dark current that is different from the bias. If you’re using a sensor that has very little dark current and you’re cooling it to maybe -20°C and below then the difference between a bias frame and maybe a 5s dark frame is not going to really be noticed. I started off with a QHY9 which also used the KAF-8300 sensor and needed at least 3-5s flat exposures to remove the shutter from the flat. In this instance I was better off with a 4s flat dark as it better represented what I wanted to remove than a bias.

Many CMOS sensors have amp glow which is another form of dark current but it is non linear, meaning, it doesn’t increase linearly as dark current does. Used both an ASI1600 and QHY163M and I’ve found (and at least QHY have suggested) that using exposures shorter than 0.3-0.5s will cause issues when it comes to calibration so the use of bias frames hasn’t been recommended.
Reply With Quote
  #8  
Old 09-11-2018, 07:36 PM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Ah! This old chestnut.

A bias is a zero second exposure duration dark.

So, if you take a dark frame longer than zero seconds, you already have the bias information.

Not all cameras have perfectly linear responses, including my not-so-cheap SBIG STX16803, hence a dark library that matches the exposure times of the light frames, is the best way to calibrate your data IMHO.

Last edited by Peter Ward; 09-11-2018 at 07:42 PM. Reason: clarification
Reply With Quote
  #9  
Old 09-11-2018, 09:03 PM
ChrisV's Avatar
ChrisV (Chris)
Registered User

ChrisV is offline
 
Join Date: Aug 2015
Location: Sydney
Posts: 1,737
Might be an old chesnut for some but not for me.

I've been calibrating my darks and flats with a master (super)bias in PI. Then calibrating my lights with these and the bias frame. Similar to this
http://www.lightvortexastronomy.com/...ixinsight.html
Writing it out now - it kind looks like a grammatically incorrect double negative?

So am I better off calibrating lights with darks and flats with dark flats. Then calibrating my lights with these calibrated darks and flats?

Sorry to hijack this Andy ...

Last edited by ChrisV; 09-11-2018 at 10:29 PM.
Reply With Quote
  #10  
Old 10-11-2018, 07:44 AM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Quote:
Originally Posted by ChrisV View Post
Might be an old chesnut for some but not for me.

I've been calibrating my darks and flats with a master (super)bias in PI. Then calibrating my lights with these and the bias frame. Similar to this
http://www.lightvortexastronomy.com/...ixinsight.html
Writing it out now - it kind looks like a grammatically incorrect double negative?

So am I better off calibrating lights with darks and flats with dark flats. Then calibrating my lights with these calibrated darks and flats?

Sorry to hijack this Andy ...
All you need are 5 to 9 or so Median combined, auto dark subtracted sky flats (one set for each filter)
and the around same number of darks (which match light frame exposure times) . Bias frames are not necessary, but can be used if don’t have darks for some odd reason.

I use MaxImCCD to calibrate data frames, but there are many alternatives.
Reply With Quote
  #11  
Old 10-11-2018, 08:11 AM
RickS's Avatar
RickS (Rick)
PI cult recruiter

RickS is offline
 
Join Date: Apr 2010
Location: Brisbane
Posts: 10,584
Quote:
Originally Posted by Peter Ward View Post
All you need are 5 to 9 or so Median combined, auto dark subtracted sky flats (one set for each filter)
With a KAF-16803 and deep wells that will give you a total integration of around 200K or more electrons and pretty good SNR. With a CMOS camera at high gain it might only give you a few thousand e- and poor SNR. It's always a good idea to do the maths:

Total electrons = number of subs x 1/2 full well (assume you're doing flats at around 30K ADU)

SNR = total e- / sqrt(total e-) = sqrt(total e-)

Somewhere around 250K e- is a good number but possibly impractical with shallow wells. It's 6 subs with my U16M and 252 subs with my ASI1600mm at a gain of 200!

Cheers,
Rick.
Reply With Quote
  #12  
Old 10-11-2018, 08:42 AM
Atmos's Avatar
Atmos (Colin)
Ultimate Noob

Atmos is offline
 
Join Date: Aug 2011
Location: Melbourne
Posts: 6,980
I usually do 25 flats and I’m generally using 12bit mode which is Gain 75-80 which gets me to about 50,000 e-.
For astrometric purposes 1,000,000 e- is recommended but if all you’re wanting to do is remove vignetting and dust motes then it isn’t that important
Reply With Quote
  #13  
Old 10-11-2018, 08:43 AM
Peter Ward's Avatar
Peter Ward
Galaxy hitchhiking guide

Peter Ward is offline
 
Join Date: Dec 2007
Location: The Shire
Posts: 8,090
Quote:
Originally Posted by RickS View Post
With a KAF-16803 and deep wells that will give you a total integration of around 200K or more electrons and pretty good SNR. With a CMOS camera at high gain it might only give you a few thousand e- and poor SNR. It's always a good idea to do the maths:

Total electrons = number of subs x 1/2 full well (assume you're doing flats at around 30K ADU)

SNR = total e- / sqrt(total e-) = sqrt(total e-)

Somewhere around 250K e- is a good number but possibly impractical with shallow wells. It's 6 subs with my U16M and 252 subs with my ASI1600mm at a gain of 200!

Cheers,
Rick.
The math on the flats is the same for your lights...so you need 250 odd subs for a decent S/N eh?
Reply With Quote
  #14  
Old 10-11-2018, 09:29 AM
Slawomir's Avatar
Slawomir (Suavi)
Registered User

Slawomir is offline
 
Join Date: Sep 2014
Location: North Queensland
Posts: 3,240
Quote:
Originally Posted by Andy01 View Post
Hi folks,

I'm using a QSI683 wsg8 and recently had a go at using Astropixel Processor. Very cool software indeed.
APP allows the user to load lights, darks, flats, dark flats, bias or masters of each of these in any combination.

Some users are recommending using bias instead of dark flats - which I understand is very convenient.
Now I'm not looking for a very technical answer, I just want to use what works best! Usually I take lights, darks, flats & dark flats, but no bias.

So my question is - should I be using Bias or Dark flats? What's best?
Bias or dark flats or both?

cheers
Andy
Andy, I think it will be best to calibrate a dataset with both methods and measure and compare the results. But I suspect that differences will be negligible. Having said that, I feel that darks would be worth trying, given that there is a good number of them in order to eliminate any external signal such as cosmic rays etc during stacking.

With my camera (ICX 814), I found that SNR decreases slightly when I use a good number of darks for calibrating up to about 100 15-minute narrowband subs, but when stacking more than 150 15-minute narrowband subs, SNR improves slightly when calibrating subs with a good number of darks. I suspect this is due to a fixed spatial camera noise. Visually, there was no noticeable difference though, no matter how hard I looked.

Given such low dark current for my camera, which only shows a just measurable improvement when using darks for 40hrs+ of data, I do not bother at all with darks for calibrating 10-20s flats (for narrowband filters), since, from what I investigated, using 50 or so darks instead of hundereds of bias frames would make master flats a tiny bit more noisy.

For calibration, I use 250 bias frames and aim for at least 600,000e for master flats.

Last edited by Slawomir; 10-11-2018 at 09:45 AM.
Reply With Quote
  #15  
Old 10-11-2018, 12:11 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 Atmos View Post
For astrometric purposes 1,000,000 e- is recommended but if all you’re wanting to do is remove vignetting and dust motes then it isn’t that important
1Me- is certainly overkill for pretty pictures but noise from a poor quality master flat will end up in your image during calibration so I reckon it's definitely worth aiming for 100K or more.

Quote:
Originally Posted by Peter Ward View Post
The math on the flats is the same for your lights...so you need 250 odd subs for a decent S/N eh?
It's better and worse than that, Peter. It's better because 250K or more electrons sets a fairly high bar for SNR - deliberately so because we don't want flat calibration to add much noise. It's worse because lights usually have data values much less than half full well in areas of interest. It is certainly true that you need to take lots of (short) subs with a CMOS camera at high gain.
Reply With Quote
  #16  
Old 10-11-2018, 12:22 PM
Andy01's Avatar
Andy01 (Andy)
My God it's full of stars

Andy01 is offline
 
Join Date: Mar 2012
Location: Melbourne
Posts: 3,253
Thanks everyone for your input & considered replies.
I’ll digest all this further but you’ve reassured me to stick with tried & true darks & dark flats with my camera.
That said, intial results using APP with lights/darks/flats & dark flats are very encouraging - now if the skies over Snake Valley will just co-operate this weekend I’ll be a very happy boy!
Reply With Quote
  #17  
Old 10-11-2018, 03:14 PM
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
With a KAF-16803 and deep wells that will give you a total integration of around 200K or more electrons and pretty good SNR. With a CMOS camera at high gain it might only give you a few thousand e- and poor SNR. It's always a good idea to do the maths:

Total electrons = number of subs x 1/2 full well (assume you're doing flats at around 30K ADU)

SNR = total e- / sqrt(total e-) = sqrt(total e-)

Somewhere around 250K e- is a good number but possibly impractical with shallow wells. It's 6 subs with my U16M and 252 subs with my ASI1600mm at a gain of 200!

Cheers,
Rick.
I have never heard this before. Interesting. I typically get 25 subs per master flat with my gear which gives me only ~ 50k. Perhaps I should set the bar a little higher here.
Reply With Quote
  #18  
Old 10-11-2018, 07:56 PM
ChrisV's Avatar
ChrisV (Chris)
Registered User

ChrisV is offline
 
Join Date: Aug 2015
Location: Sydney
Posts: 1,737
Thanks for asking about this Andy. I think I might have learnt something here ... Possibly ....
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 08:07 AM.

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