#1  
Old 12-05-2008, 04:01 AM
Ingo
Registered User

Ingo is offline
 
Join Date: Apr 2007
Location: Albany, NY
Posts: 559
ST8 Blue Problem

I am using an SBIG ST8 and I cannot get blue to come out right.

I did

R: 120s
G: 120s
B: 240s

And my M20 Trifid came out yellow instead of pinkish red =(

Any help?
Reply With Quote
  #2  
Old 12-05-2008, 04:43 AM
gbeal
Registered User

gbeal is offline
 
Join Date: Oct 2004
Location: New Zealand
Posts: 4,345
No way am an expert on this, but I know someone who is and he will reply once he sees this. Why the difference in exposure times. What filters as well?
Gary
Reply With Quote
  #3  
Old 12-05-2008, 04:49 AM
Ingo
Registered User

Ingo is offline
 
Join Date: Apr 2007
Location: Albany, NY
Posts: 559
The different exposure time is because I head somebody say blue is half sensitive as red and green.
Reply With Quote
  #4  
Old 12-05-2008, 05:05 AM
iceman's Avatar
iceman (Mike)
Sir Post a Lot!

iceman is offline
 
Join Date: Sep 2004
Location: Gosford, NSW, Australia
Posts: 36,760
Your exposure times should be the same with a monochrome camera + filters.

Can you post the image so we can see what it looks like?

What filters are you using? How are you combining the RGB data?
Reply With Quote
  #5  
Old 12-05-2008, 05:13 AM
Ingo
Registered User

Ingo is offline
 
Join Date: Apr 2007
Location: Albany, NY
Posts: 559
Quote:
Originally Posted by iceman View Post
Your exposure times should be the same with a monochrome camera + filters.

Can you post the image so we can see what it looks like?

What filters are you using? How are you combining the RGB data?
I'm combining the RGB by photoshop channels.
Attached Thumbnails
Click for full-size image (m20.jpg)
186.8 KB44 views
Reply With Quote
  #6  
Old 12-05-2008, 05:41 AM
iceman's Avatar
iceman (Mike)
Sir Post a Lot!

iceman is offline
 
Join Date: Sep 2004
Location: Gosford, NSW, Australia
Posts: 36,760
Quote:
Originally Posted by iceman View Post
What filters are you using?
...
Reply With Quote
  #7  
Old 12-05-2008, 06:55 AM
sjastro's Avatar
sjastro
Registered User

sjastro is offline
 
Join Date: Jun 2007
Posts: 2,926
If you are using a "non E" version ST-8 CCD and the original SBIG filter set the RGB ratios are approximately 1:1.5:3.

What you should do however is calculate the ratios using a reference G5 star.

Regards

Steven
Reply With Quote
  #8  
Old 12-05-2008, 08:18 AM
jase (Jason)
Registered User

jase is offline
 
Join Date: Sep 2006
Location: Melbourne, Victoria
Posts: 3,916
Ingo,
If your neb turned out yellow, adding more blue is only going to turn it green. Sounds like your colour balance ratios are out for the other channels too. Based on your exposure lengths, you are using a basic 1:1:2 ratio. As Steven points out, the ratio is likely to be higher in the blue, so you’ll need to increase the length of your blue exposures. Keep an eye on the R and G channels as well. I doubt they are 1:1 respectively.

If you are planning on using exposure lengths for the method of balancing images, keep in mind that this method is dependant on S/N ratio. i.e the channel (say blue) will have a high signal that the R and G channels for the image to colour balance correctly. While this sounds relatively simple in theory, I’ve struggled to balance images well using this technique. The problem is…when you combine your individual R, G, B exposures, the combine method you use will naturally alter the S/N. The only time I could get this method to work was using the a Sum combine – which in all honesty…is crap, because it does little in the way of noise reduction. Median or Sigma Reject/clip is a far better options, but these alter the S/N.

By all means try the exposure length method, but you may wish to start with taking equal exposures through the individual filters, then colour combine them using different ratios.

To find your correct colour balance based on a G2V star, follow these steps. There are a few ways of calculating your colour balance for a specific imaging train. Ideally, you should perform the calibration through each optical system (telescopes, lenses etc) that you intended to use. Don’t expect the weighting for one optical system to be the same for others – even if you’re using the same CCD camera and filters!

There are many resources on the web that tell you how to do perform the calibration and there is more than one-way of doing it. Here’s how I do it. It may seem a little daunting at first, but once you’ve done it a couple of times it’s straight forward.
  • Firstly, look up a table of G2V stars and select one close to zenith. You want to minimise as much extinction factor as possible. Also choose a star that isn’t going to saturate easily. There is a great reference table here - http://www.skymap.com/g2v.htm
  • Once you’ve selected your target star, begin to focus the star as best you can through your optical system. Proceed to take 10 or more equal exposures through each individual filter (R, G, B). I generally go for 30 second exposures (don't forget to dark,bias,flat process them!).
  • Register and average or median (combine) the frames.
  • In MaximDL or other astro imaging software, I then measure the intensity of the G2V star in each frame (R, G, B). Make sure you’ve selected the right star!!! This sounds stupid, but when you’ve got a huge FOV you may need to reference your astro planning software or star charts. You need to ensure the measurement encompass the entire star, not just the core.
  • Take note of what the intensity values are;
As an example:
R = 600,000
G = 700,000
B = 500,000
  • So based on these (example – don’t use these!) figures, you can see the example optical system is most sensitive in the green wavelengths. What ever is the most sensitive channel, make that ratio value of 1. So now that we have established G = 1, we can compute the R and B weights. If you divide the green number (700,000) by the red number (600,000) you get 1.16. Divide the green number by the blue number you get 1.4.
  • So your calibrated RGB weightings are R:1.6, G:1,B:1.4 for that specific optical system/filter/camera combination.
From this point on, you can take equal exposure lengths as you image certain objects and use the correct weightings for each channel to obtain the right balance.

You should perform this even if you’re using true-balance filters such as those made by Astrodon. You’ll find the spectral response of your entire optical train will not deliver a RGB of 1:1:1. There are likely to be subtle variations. My method above method can be also used for colour imaging with UVBRI filters – in particular BVR (not so much U and I as they are out of the visual spectrum).

Finally, make sure you "normalise" you individual combined R,G,B files. If the average background ADU is brighter in one of the channels, the image will not balance well. MaximDL has a automatic normalisation feature, though you usually get better results normalising manually. Analyse each of the R,G,B combined files - look at the darkest area (same location) in each image, then use pixel math to subtract a determined value from each of the images so that the average background ADU is equal (or close enough) to 50 (others may use a different value, but 50 works ok for me). The component images should then balance correctly. Make sure you don't measure the average background ADU where there is nebulosity or a gradient is present as this will through out your figures. Sample a few corners of the image usually works.

A little pain for long term gain... Welcome to the world of monochrome imaging with R,G,B filters! Those OSC guys have it too easy!

UPDATE: I've attached a DB search dump from TheSky listing the G2V stars - enjoy
Attached Files
File Type: txt g2v.txt (60.0 KB, 17 views)

Last edited by jase; 12-05-2008 at 09:20 PM. Reason: Just realised the folks at Gemini.edu changed their G2V page... sorry about that.
Reply With Quote
  #9  
Old 13-05-2008, 08:54 AM
Ingo
Registered User

Ingo is offline
 
Join Date: Apr 2007
Location: Albany, NY
Posts: 559
I think I'm using too much green. I think it might be 1:0.5:2

Reply With Quote
  #10  
Old 13-05-2008, 11:33 AM
jase (Jason)
Registered User

jase is offline
 
Join Date: Sep 2006
Location: Melbourne, Victoria
Posts: 3,916
Quote:
Originally Posted by Ingo View Post
I think I'm using too much green. I think it might be 1:0.5:2

No idea Ingo. You're on your own until you can advise what filters you are using as they all respond differently to spectral wavelengths. Also, need to know if the cam is an ST8E or ST8XME. Once known, I'm happy to respond in an well informed manner, instead of guessing. You've now got the information to perform a G2V star test to reach the correct channel weightings.
Reply With Quote
  #11  
Old 13-05-2008, 11:58 AM
rogerg's Avatar
rogerg (Roger)
Registered User

rogerg is offline
 
Join Date: Feb 2006
Location: Perth, Western Australia
Posts: 4,563
Jace,

If I understand you correctly, then providing the background levels are normalised between the exposures as you say (eg to 50), there is no need to know the exact relationship between the filters? ... because whatever the relationship is, they're all now equal.

Thanks,
Roger.
Reply With Quote
  #12  
Old 13-05-2008, 12:54 PM
montewilson's Avatar
montewilson (Monte)
Registered User

montewilson is offline
 
Join Date: Apr 2007
Location: Sydney
Posts: 374
Although it is unlikely I thought I should at least ask...

Are you sure the filters are in the correct spots? You'll never win if they aren't.
Reply With Quote
  #13  
Old 13-05-2008, 02:08 PM
jase (Jason)
Registered User

jase is offline
 
Join Date: Sep 2006
Location: Melbourne, Victoria
Posts: 3,916
Quote:
Originally Posted by rogerg View Post
Jace,

If I understand you correctly, then providing the background levels are normalised between the exposures as you say (eg to 50), there is no need to know the exact relationship between the filters? ... because whatever the relationship is, they're all now equal.

Thanks,
Roger.
No, normalising (aka equalising) the individual RGB component images does not alter the relationship between the filters i.e. it is not trying to compensate for filter transmission characteristics or CCD spectral response.
What normalising the background level does is provide consistency across all component RGB images allowing the colour weight ratios to be true (as validated via G2V calibration).

If you don't normalised the individual RGB component images before colour combining, you need to alter your colour weights to compensate for the imbalance. You need to know how much you need to change the colour weights to compensate for the indifference.

Lets take an example (rather basic one, but could be more complex i.e across multiple filters or environmental conditions i.e. light polution). You're using Astrodon 1:1:1 Tru-colour balance filters. As the moon rises it alters the light intensity of the blue filter, hence when you perform a colour combine of 1:1:1 ratios, your image has a blue shift. If you measure the background ADU of the blue channel and compare it to the R and G, it will be undoubtedly different and the cause of the issue. Now you can do one of two things, alter the weights to tone down the blue (of which you still need to measure the background level of the images to calculate the difference) or use pixel math to normalise all component image and continue to use the 1:1:1 ratios. The latter will give you the greater accuracy - in my experience.

Software like MaximDL performs an automatic normalisation function. I feel certain CCDSoft would have similar. Be careful using the automated functions if your image has nebulosity throughout i.e. there is technically no "real" background. In such cases, the best way of getting accurate colour images is to normalise manually. There is no free cut lunch if you want the accuracy - you need to work for the result.
Reply With Quote
  #14  
Old 13-05-2008, 02:41 PM
rogerg's Avatar
rogerg (Roger)
Registered User

rogerg is offline
 
Join Date: Feb 2006
Location: Perth, Western Australia
Posts: 4,563
Thanks Jase. Never have been sure how to ACCURATELY equate a known imbalance in filters into actual image adjustments. I have completely overlooked the CCDSoft LRGB processing since early attempts gave ugly results and henceI turned to PS. I see from web searches it has the ratio adjustment, and normalize exposures and normalize backgrounds. I'll give it another shot on some old data perhaps (haven't collected any RGB for months now).

PS. Anyone interested in doing this with CCDSoft or Maxim I would suggest trying google with the term "ccdsoft "Normalize Backgrounds"". You should find a few documents which give extensive information and wonderful help, but I suspect should not be un-secured or indexed by Google. However, they are, not our fault!

Roger.
Reply With Quote
  #15  
Old 13-05-2008, 03:08 PM
jase (Jason)
Registered User

jase is offline
 
Join Date: Sep 2006
Location: Melbourne, Victoria
Posts: 3,916
Quote:
Originally Posted by montewilson View Post
Are you sure the filters are in the correct spots? You'll never win if they aren't.
Actually, its not as stupid sounds. I have got caught out. The problem is certain software references filter wheel slot numbers differently. Some start at 0 and others 1. Thought I was going crazy until I RTFM
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 06:20 PM.

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