#1  
Old 04-02-2010, 12:10 PM
Hagar (Doug)
Registered User

Hagar is offline
 
Join Date: Oct 2006
Location: Australia
Posts: 4,646
Question Latest ramblings on Flat files.

I have recently been doing a bit of work with my QHY8Pro with regard to flat frame images.
You may already be aware of all this but it may well be a different approach to the flat file problems a lot of us suffer.

With a lot of the new cameras having a well depth of only 25ke, I was under the impression my fats should have a maximum ke value of between a third and half the well depth. I often found these flats to be near useless.

Sitting back and thinking about the methods used to capture flats and in particular the software used I came to the conclusion (right or wrong) that the process used by me was wrong.

The image is taken with a camera with a well depth of say 25ke. It runs the image through the cameras analogue to digital converter and down loads the image to the laptop at full 16 bit depth. At this bit depth it transforms the image into a format which makes it appear as a well depth of 65000e or, well you all know the figure anyway. When measurements are taken at the capture software on the computer which indicates a full 65000e.

The assumption I have made is that when using a CCD with an AtoD converter our flats should be taken between a third and half the resultant output as at the computer. ie. 65ke So in essence instead of capturing flats at 9 to 12 thousand we should be using ADU figures between 21000 and 32000e.

The few experiments I have done do seem to indicate much better results from flats taken at this much higher exposure and ADU count.

I am interested to hear if anyone else has found the same improved results or has noticed any other ways to improve the use of flat files.
Reply With Quote
  #2  
Old 04-02-2010, 01:39 PM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Sydney
Posts: 22,060
Hi Doug when you say 65ke I assume you mean 65k ADU right? My understanding is that a third of the full well is recommended, the full well being the value at which the camera sensor is starting to have a non linear response which in my QHY8 appears around 45kADU. In practice I shoot flats at 9~10kADU on the brightest pixels of the bayer matrix which with my EL sheet ends up in the red pixels. I tried higher values but anything over 10kADU is giving me reverse vignetting no matter how I scale/normalise to my lights.
Reply With Quote
  #3  
Old 04-02-2010, 02:46 PM
Terry B's Avatar
Terry B
Country living & viewing

Terry B is offline
 
Join Date: Mar 2006
Location: Armidale
Posts: 2,789
Doug
Flats will introduce noise. To limit this noise the level of the flat needs to be as high as possible without saturating or in the case of CCD with antiblooming, as high possible within the linear response range of the CCD.
To measure this you need to know when the linear part of the CCD ends.
You can test this by taking increasing exposures of a constant light source and measuring the peak response. Graph this against the exposure time and see where it becomes non linear.
The constant light source can be a non variable star at the zenith.
You then expose your flat so that the brightest part is just in the linear part of the CCD.
Reply With Quote
  #4  
Old 04-02-2010, 02:51 PM
Hagar (Doug)
Registered User

Hagar is offline
 
Join Date: Oct 2006
Location: Australia
Posts: 4,646
Quote:
Originally Posted by multiweb View Post
Hi Doug when you say 65ke I assume you mean 65k ADU right? My understanding is that a third of the full well is recommended, the full well being the value at which the camera sensor is starting to have a non linear response which in my QHY8 appears around 45kADU. In practice I shoot flats at 9~10kADU on the brightest pixels of the bayer matrix which with my EL sheet ends up in the red pixels. I tried higher values but anything over 10kADU is giving me reverse vignetting no matter how I scale/normalise to my lights.
Hi Marc, Thanks for the reply. I also use an EL sheet which has made taking flats a bit easier but my results have been mixed.
I had always done my Flats at about the 9000 to 12000 range and had some very mixed results. Sometimes good, sometimes bad. With the RC having such a long imaging train it does suffer quite a lot from Vignetting and the old settings just didn't cut the mustard on this image. Using a max of 22000 produced much better results. I actually way oversaturaded the original image without flat calibration which showed up the problem areas quite graphically. I then re-did the image after calibration and again way oversaturated the image to find the 22k flats had really carried out the desired job.
I tried this at 7000, 9000,12000,15000 and 22000 and was surprised by the results. The 22000 calibration was well out in front. I will try this on some other targets when the sky clears and I can get some other images. The only scope I can use this test method on is the RC as the FSQ has a much larger image circle and doesn't suffer from any vignetting with the QHY8 CCD.

Just my ramblings of course but I thought it was worth sharing. Others may have experienced the same unreliable application of flats that I have in the past.
Reply With Quote
  #5  
Old 04-02-2010, 03:04 PM
Hagar (Doug)
Registered User

Hagar is offline
 
Join Date: Oct 2006
Location: Australia
Posts: 4,646
Quote:
Originally Posted by Terry B View Post
Doug
Flats will introduce noise. To limit this noise the level of the flat needs to be as high as possible without saturating or in the case of CCD with antiblooming, as high possible within the linear response range of the CCD.
To measure this you need to know when the linear part of the CCD ends.
You can test this by taking increasing exposures of a constant light source and measuring the peak response. Graph this against the exposure time and see where it becomes non linear.
The constant light source can be a non variable star at the zenith.
You then expose your flat so that the brightest part is just in the linear part of the CCD.
Thanks Terry, something else to try and see if I can get my head around it. As i said I really don't have any problems with the FSQ but find I do need to work on something better using the little RC.
Reply With Quote
  #6  
Old 04-02-2010, 03:51 PM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Sydney
Posts: 22,060
Quote:
Originally Posted by Hagar View Post
I had always done my Flats at about the 9000 to 12000 range and had some very mixed results. Sometimes good, sometimes bad. With the RC having such a long imaging train it does suffer quite a lot from Vignetting and the old settings just didn't cut the mustard on this image. Using a max of 22000 produced much better results.
That's interesting. I've always struggled to get correct flats on the C11 at F/10. I get them spot on with the 5" newt at F/5 with 9000ADU. Text book flat fielding - dust motes and all vigneting gone. But when I do the same levels on the C11 it's not quite there. Might try to increase the ADU see what happens. Maybe the FL does effect the illumination repartition on the CCD somehow. Will have to try next imaging session if those damn clouds want to go away that is.
Reply With Quote
  #7  
Old 04-02-2010, 06:12 PM
allan gould's Avatar
allan gould
Registered User

allan gould is offline
 
Join Date: May 2005
Location: Brisbane
Posts: 4,485
Really interesting post Doug. As with what Mark stated my flats were great with my 127mm scope but when I shifted to the 10"sct they ate not adequAte. I wiLl have to do a range of flats to see whAt works best with this scope.
Reply With Quote
  #8  
Old 04-02-2010, 06:44 PM
Bassnut's Avatar
Bassnut (Fred)
Narrowfield rules!

Bassnut is offline
 
Join Date: Nov 2006
Location: Torquay
Posts: 5,064
Quote:
Originally Posted by Hagar View Post
Hi Marc, Thanks for the reply. I also use an EL sheet which has made taking flats a bit easier but my results have been mixed.
I had always done my Flats at about the 9000 to 12000 range and had some very mixed results. Sometimes good, sometimes bad. With the RC having such a long imaging train it does suffer quite a lot from Vignetting and the old settings just didn't cut the mustard on this image. Using a max of 22000 produced much better results. I actually way oversaturaded the original image without flat calibration which showed up the problem areas quite graphically. I then re-did the image after calibration and again way oversaturated the image to find the 22k flats had really carried out the desired job.
I tried this at 7000, 9000,12000,15000 and 22000 and was surprised by the results. The 22000 calibration was well out in front. I will try this on some other targets when the sky clears and I can get some other images. The only scope I can use this test method on is the RC as the FSQ has a much larger image circle and doesn't suffer from any vignetting with the QHY8 CCD.

Just my ramblings of course but I thought it was worth sharing. Others may have experienced the same unreliable application of flats that I have in the past.
What do you mean by "better results", if you have vignetting, then the flats should show this, so that it can help fix it in the light exposure. The more vignetting, dust motes etc showing in the flat the better, its supposed to image the faults. The worse flats you can get is a uniform grey with no imperfections, thats useless, waste of iime, and of no "correction" value at all.
Reply With Quote
  #9  
Old 04-02-2010, 06:49 PM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Sydney
Posts: 22,060
Quote:
Originally Posted by Bassnut View Post
What do you mean by "better results", if you have vignetting, then the flats should show this, so that it can help fix it in the light exposure. The more vignetting, dust motes etc showing in the flat the better, its supposed to image the faults. The worse flats you can get is a uniform grey with no imperfections, thats useless, waste of iime, and of no "correction" value at all.
I think what Doug means is that if your flat is underexposed then by the time you subtract the bias and divide it in your light it will overcorrect so pushing the ADU in your flat yields different results. Of course if you push them too hard then you lose all the gritty details and dust but I still think the sweet spot is not the same for different setups and FL/aperture combinations. Otherwise the levels we tried would work for everything and they're not?
Reply With Quote
  #10  
Old 04-02-2010, 07:18 PM
ejcruz's Avatar
ejcruz (Eddie)
Registered User

ejcruz is offline
 
Join Date: Nov 2008
Location: Toongabbie, NSW
Posts: 140
I shoot all my flats between 22K-23K for both QHY8 and QHY9 using either WO FLT110 or RC8, 3 sec exposure time, tried below 22K and got mixed results.

Cheers
Eddie
Reply With Quote
  #11  
Old 04-02-2010, 07:21 PM
Hagar (Doug)
Registered User

Hagar is offline
 
Join Date: Oct 2006
Location: Australia
Posts: 4,646
Quote:
Originally Posted by Bassnut View Post
What do you mean by "better results", if you have vignetting, then the flats should show this, so that it can help fix it in the light exposure. The more vignetting, dust motes etc showing in the flat the better, its supposed to image the faults. The worse flats you can get is a uniform grey with no imperfections, thats useless, waste of iime, and of no "correction" value at all.
Quote:
Originally Posted by multiweb View Post
I think what Doug means is that if your flat is underexposed then by the time you subtract the bias and divide it in your light it will overcorrect so pushing the ADU in your flat yields different results. Of course if you push them too hard then you lose all the gritty details and dust but I still think the sweet spot is not the same for different setups and FL/aperture combinations. Otherwise the levels we tried would work for everything and they're not?
Thanks Mark, That pretty much covers it. It is difficult to explain properly, particularly considering my limited experience with flats of late.
They seem to be a fairly fiddly item to image correctly so as to get the very best out of your image and not just the quick fix.
Reply With Quote
  #12  
Old 06-02-2010, 10:49 AM
Paul Haese's Avatar
Paul Haese
Registered User

Paul Haese is offline
 
Join Date: Jan 2009
Location: Adelaide
Posts: 9,937
Quote:
Originally Posted by Bassnut View Post
What do you mean by "better results", if you have vignetting, then the flats should show this, so that it can help fix it in the light exposure. The more vignetting, dust motes etc showing in the flat the better, its supposed to image the faults. The worse flats you can get is a uniform grey with no imperfections, thats useless, waste of iime, and of no "correction" value at all.
That is my understanding too. I usually image for each colour until I can see the central dust motes for that colour; this also shows vignetting really well. It varies quite a lot really for each filter with exposure. I suppose I should use a numbers relationship but I go by what I can see in the flat image. This usually gives me a good flat to subtract away from the subs.
Reply With Quote
  #13  
Old 06-02-2010, 04:02 PM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Sydney
Posts: 22,060
Quote:
Originally Posted by Paul Haese View Post
That is my understanding too. I usually image for each colour until I can see the central dust motes for that colour; this also shows vignetting really well. It varies quite a lot really for each filter with exposure. I suppose I should use a numbers relationship but I go by what I can see in the flat image. This usually gives me a good flat to subtract away from the subs.
Hi Paul, are you using a monochrome camera?
Reply With Quote
  #14  
Old 06-02-2010, 05:41 PM
Terry B's Avatar
Terry B
Country living & viewing

Terry B is offline
 
Join Date: Mar 2006
Location: Armidale
Posts: 2,789
I try to have very good flats because I'm using my images for photometry. I take flats through each filter separately as there is a little bit of difference in each. It is very hard to get perfect flats.
You can check the flats by applying them to an image that is basically just a star field. ie no nebula and then stretching the histogram to exagerate the background. It should be perfectly even across the entire field.
The image I have attached is a V filter image of the nova KT Eri.
It is stretched to show only values between 260 and 330 in a 16 bit image (ie 0 to 65000)
It was a 30 sec binned image 2x2 from my ST10XME and I have shrunk it 50% to display here.
It is not perfect after flat fielding. There is still a little bit of uneven illumination in the left corners but it shows the effect of the flat well.
Attached Thumbnails
Click for full-size image (KTERI.jpg)
118.8 KB23 views
Click for full-size image (KTERI-raw.jpg)
112.6 KB21 views
Reply With Quote
  #15  
Old 07-02-2010, 05:47 PM
Paul Haese's Avatar
Paul Haese
Registered User

Paul Haese is offline
 
Join Date: Jan 2009
Location: Adelaide
Posts: 9,937
Quote:
Originally Posted by multiweb View Post
Hi Paul, are you using a monochrome camera?
Yes Marc. I am using a QSI583 monochrome.
Reply With Quote
  #16  
Old 07-02-2010, 05:54 PM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Sydney
Posts: 22,060
Quote:
Originally Posted by Paul Haese View Post
Yes Marc. I am using a QSI583 monochrome.
You've made some interesting comments and I'm starting to think your method of accessing the flats levels per color channel might also be relevant to an OSC holding a bayer matrix. Might have to shoot three set of flats one for each channel, still flat field prior to debayer but then use the proper flat intensity for the proper channel and discard the other two.
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 09:32 PM.

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