Background Neutralization and Color Calibration

[rdryfoos]

I am having trouble using Background Neutralization on images that have little dark background due to the presence of nebulosity.  If there is no suitable background sample, where should I put my preview to use in the BN tool?  Not having a decent background disrupts the image by increasing the lightness of the background which throws the Color Calib. off as well.  It has always seemed more "true" to identify a real white point based on the filters,  camera, and calibration stars (typically GV2)  That way--with the lack of a background sample will not be that relevant.

Short of that--how to use Background Neutralization if there is no background?

Thanks   Rodd

[Cabissi]

Rodd,

I am having a similar problem with my image of the Eastern Veil nebula, However my problem is not too much nebula, but rather toooooo many stars.  The only areas clear of stars are very tiny and don't seem to work well.  Is then any way to link several previews to get a statistically legitimate background sample?

Rodd, is there any region of dark nebula that you can sample?

Chris

[rdryfoos]

Rodd,

I am having a similar problem with my image of the Eastern Veil nebula, However my problem is not too much nebula, but rather toooooo many stars.  The only areas clear of stars are very tiny and don't seem to work well.  Is then any way to link several previews to get a statistically legitimate background sample?

Rodd, is there any region of dark nebula that you can sample?

Chris


Chris--No, no dark nebula.  I do believe you can take the average of multiple previews for your situation.  Don't know the specifics of how to do it, but I remember reading it somewhere.

Rodd

[pfile]

you can define a bunch of previews on the image and then use the script called PreviewAggregator to make a new image composed of just those previews. then in BN/CC you need to specify that image as the reference instead of the default (target image).

rob

[rdryfoos]

you can define a bunch of previews on the image and then use the script called PreviewAggregator to make a new image composed of just those previews. then in BN/CC you need to specify that image as the reference instead of the default (target image).

rob

What about my problem pfile?  How to establish a neutralized background when there is no background to select?

[pfile]

dunno - it's been suggested by vicent that if you have this problem then you need to shoot another target at the same alt/az, hopefully at the same time of the night (which means a little earlier or later in the year, but not too much) which does not have nebulosity in the background and use that as the background reference/color calibration image...

rob

[rdryfoos]

dunno - it's been suggested by vicent that if you have this problem then you need to shoot another target at the same alt/az, hopefully at the same time of the night (which means a little earlier or later in the year, but not too much) which does not have nebulosity in the background and use that as the background reference/color calibration image...

rob
2 things--I can't be the only one with this issue.  What do people do?.  Vicent is very adamant and proud of PI for not using the GV2 white point (objective) color calibration.  He says the PI approach of color calibration being relative within an image is best.  Same is true for background.  PI must have an answer to this question.  and 2), So it is possible to use a preview from another image for background Neutralization?  I can just choose an image of a dark sky and use that?

[pfile]

no, you can't use any random image, because the color of the background in your image is going to be most strongly influenced by the altitude of the object and the local light pollution conditions, etc. that is why you want to make your reference under similar conditions as your real image. i don't know what object you're shooting but chances are there's a patch of sky nearby that is not filled with nebula?

i think the g2v thing is a red herring... if there's a bunch of dust/gas between you and the reference stars, or if the image was subject to atmospheric extinction, the G2V calibration's going to be wrong anyway. if you get a reference image that's free of these problems then G2V or the PI method should yield similar results.

also you need to write your replies outside of the

Code: [Select]

[quote][/quote] tags or else they are pretty impossible to read..

rob

[rdryfoos]

no, you can't use any random image, because the color of the background in your image is going to be most strongly influenced by the altitude of the object and the local light pollution conditions, etc. that is why you want to make your reference under similar conditions as your real image. i don't know what object you're shooting but chances are there's a patch of sky nearby that is not filled with nebula?

i think the g2v thing is a red herring... if there's a bunch of dust/gas between you and the reference stars, or if the image was subject to atmospheric extinction, the G2V calibration's going to be wrong anyway. if you get a reference image that's free of these problems then G2V or the PI method should yield similar results.

also you need to write your replies outside of the

Code: [Select]

[quote][/quote] tags or else they are pretty impossible to read..

rob

  Not sure if this is what you meant by outside the code.  We'll see.  When I use my 1960 FL scope (C11 Edge with .7x reducer) the FOV is quite small.  I have tried making the preview really tiny--but in some images, there may be background color.   I have used BE and have had the back ground get lighter after its use, not darker.  In those instances, can one do CC without BE?  This happened to me with my 540 MM FL refractor as well.  There was no black background and the use of BN made the image lighter.  Short of clipping the Hist, what to do?  Masking and using curves did not work because the background was partially covered by teh mask do to the brightness of the background.  Also--when I say G2V--I mean a standardized value--not an actual star.  There is a standard G2V color and that could be used, and coloration would be based on a scale using that as a benchmark--not a star you select in the FOV.  Knowing the G2V value, the filter type and the camera type, and maybe the declination of the object if you want to get picky, intervening dust is removed from the equation.

[pfile]

using a code block was the only way i knew to get the quote tags to show up. but it looks like you got it.

i suppose you can use color calibration without BN, but i'm not sure the results would be accurate.

much has been written about the G2V vs. pixinsight method of color calibration and i don't think i can add any more. G2V calibration requires that you find stars in your FOV that are, well, class G2V, so you do need actual stars in order to accomplish the calibration... because what you're doing is adjusting the color balance until that (or those) G2V stars in your FOV look white.

anyhow i can't really think of anything to recommend other than trying to get an image under similar conditions which is near your target and use that for the background reference.

rob

[rdryfoos]

using a code block was the only way i knew to get the quote tags to show up. but it looks like you got it.

i suppose you can use color calibration without BN, but i'm not sure the results would be accurate.

much has been written about the G2V vs. pixinsight method of color calibration and i don't think i can add any more. G2V calibration requires that you find stars in your FOV that are, well, class G2V, so you do need actual stars in order to accomplish the calibration... because what you're doing is adjusting the color balance until that (or those) G2V stars in your FOV look white.

anyhow i can't really think of anything to recommend other than trying to get an image under similar conditions which is near your target and use that for the background reference.

rob

  Are you implying that you have never taken an image that is all nebulosity without a nebulous free background?  How about a close in shot of M42--or the Pacman Nebula, or the North American Nebula, or Thor's Helmet....I could go on and on.   There are thousands of examples.  This is a common occurrence and many many people must face it.  What do they do?  They must do something. How about a close in of M51, or M31 that fills the FOV.  I've seen many images like this.  Many were processed in Pixinsight. 

[jkmorse]

Rodd,

In situations like you describe, if you are not willing to go the extra step suggested by Rob, then the only real option is to try and align the histograms of the three color channels.  Easiest way to do that is to use Histogram Transform.  What you need to achieve, since you are looking to balance the background, is to make sure that the left/dark side of the three color channels are aligned.  If they are not, you can tweak each color by selecting it and making small adjustments until they are balanced on the left side.  That should give you an even background and help balance the image generally but there are no guarantees that this is the right answer when you are dealing with heavy and pervasive nebulosity.  What is does do is make sure that what little true background may be available in the niches is balanced. 

Another option, though a bit more heavy handed, is to use Linear Fit before combining the color stacks and simply force all three stacks onto the same curve, then combine them.

At the end of the day, you are going to have to eyeball it, seeing if what you end up with makes sense.  No program can give you a perfectly balanced background where there is no background to work with in the first instance. 

Best,

Jim 

[rdryfoos]

Rodd,

In situations like you describe, if you are not willing to go the extra step suggested by Rob, then the only real option is to try and align the histograms of the three color channels.  Easiest way to do that is to use Histogram Transform.  What you need to achieve, since you are looking to balance the background, is to make sure that the left/dark side of the three color channels are aligned.  If they are not, you can tweak each color by selecting it and making small adjustments until they are balanced on the left side.  That should give you an even background and help balance the image generally but there are no guarantees that this is the right answer when you are dealing with heavy and pervasive nebulosity.  What is does do is make sure that what little true background may be available in the niches is balanced. 

Another option, though a bit more heavy handed, is to use Linear Fit before combining the color stacks and simply force all three stacks onto the same curve, then combine them.

At the end of the day, you are going to have to eyeball it, seeing if what you end up with makes sense.  No program can give you a perfectly balanced background where there is no background to work with in the first instance. 

Best,

Jim

  Thanks Jim--but the question is  "what do you (meaning anyone whom answers) do?  It seems as if I am asking a rare question, like no body runs into this problem.  Since Background Neutralization must be done on a color image--I know of no-one that has shot a complete image  RGB or NB with the same iteration times as the main image just to get a background neutralization.  Have you ever done that?  Shout an extra 24 hours worth of data of blank sky just to neutralize teh background of your image.  I think not.   Why is it that I seem to be the only person that has run into this problem?

[msmythers]

Just thought I'd chime in. Why would you need to shoot an extra 24 hours worth of data? I would think even short exposure would give the information for BN to sample from. If your image has no neutral background then I would assume it's not a wide field. Even swinging a few minutes or a degree away to an area with a neutral background for a short exposure I would think would work.


Mike

[rdryfoos]

Just thought I'd chime in. Why would you need to shoot an extra 24 hours worth of data? I would think even short exposure would give the information for BN to sample from. If your image has no neutral background then I would assume it's not a wide field. Even swinging a few minutes or a degree away to an area with a neutral background for a short exposure I would think would work.


Mike

  If I process a color image with 20 hours of data, then the image I use for my background sample should have an equivalent amount of data (or there about) no?  Background will definitely be different in a RGB (or NB) image containing 20 minutes of data compared to an image containing 20 hours of data--maybe not from the moon, but from my back yard--definitely.  So--might as well assign a value for background--which was the whole point.