How to set light po...
 
Share:

How to set light pollution boxes correctly  

  RSS

(@schurig)
Red Giant Customer
Joined: 2 years ago
Posts: 65
March 13, 2019 12:41  

Hi,

I apologize if this has been asked before, tried to search older topics but could not find a suitable match.

So, I want to know

a) how do I place my boxes in the light pollution tool correctly:
- only in areas where there is no light pollution or minimal light pollution?
- only in areas where there is significant light pollution?
- overall the entire image

I do understand, that I don't have to place my boxes over objects or nebulosity, but what actually is the exact information I indicate to APP when placing a box?

b) I often have the case, that the light pollution with green filter is stronger than with red filter. Is it necessary to use the tool for each channel seperately before combing channels or does the light pollution algorithm take care of each channel in a composite image anyway?

Kind regards, Stephan

This topic was modified 4 months ago by schurig
This topic was modified 3 months ago by Mabula Haverkamp - Admin

ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 2 years ago
Posts: 2081
March 14, 2019 13:13  

Hi Stephan @schurig,

Have you watched this extensive tutorial about Light Pollution correction? It should answer a lot of your questions 😉

https://www.astropixelprocessor.com/part-4-light-pollution-correction-and-background-calibration-of-the-mosaic-tutorial-milky-way-to-rho-ophiuchi-by-mabula/

a) how do I place my boxes in the light pollution tool correctly:
- only in areas where there is no light pollution or minimal light pollution?

You actually don't know a lot about the amount of Light Pollution (LP) in a certain area when you start the correction. And if there is little, or a lot, makes no difference in creating a correct correction model. Any data given to the tool that should only contain sky background and stars, (stars are no problem in the selected areas 😉 ) and NO nebulosity, is what the purpose is to be able to calculate a correction model.

- only in areas where there is significant light pollution?

So no.. to be able to create a correction model, areas need to be selected where there is no nebulosity or surrounding glow of a galaxy for instance. Each box that you place, being on a spot with a lot, or a little amount of LP, will help in creating the correction model. We need to get samples of areas where there is only sky background and stars 😉

- overall the entire image

It's best to place the boxes uniformly over the entire image if possible. Don't place too many boxes, only when a small amount of boxes is not adequate for the data. Clustering of a lot of little boxes is normally not needed as you can see in my video tutorial.

I do understand, that I don't have to place my boxes over objects or nebulosity, but what actually is the exact information I indicate to APP when placing a box?

We need to get samples of areas where there should only be sky background and stars 😉

b) I often have the case, that the light pollution with green filter is stronger than with red filter. Is it necessary to use the tool for each channel seperately before combing channels or does the light pollution algorithm take care of each channel in a composite image anyway?

The color channels are treated separately, so Red, Green, Blue get their own correction 😉 So, it is much easier and better to correct the color image than the separate layers yourself. Only when correcting the color image, you can be assured that the sky backgorund actually is gray/neutral because you can check it visually.

If you try to correct the separate channels and then combine them, you will most likely see, that it's not good... So you actually need to correct on the color image to be able to see if the correction is good 😉

A tip, also shown in the video tutorial: while working in the correction tool:

  • stretch strongly with the preview filter !,
  • with an extreme amount of saturation
  • without saturation protection for the background

That will help you a lot in seeing what needs to be corrected and if the correction is adequate 😉

Kind regards,

Mabula

This post was modified 3 months ago 2 times by Mabula Haverkamp - Admin

Main developer of Astro Pixel Processor and owner of Aries Productions


ReplyQuote
(@schurig)
Red Giant Customer
Joined: 2 years ago
Posts: 65
March 14, 2019 13:22  

Hi Mabula,

Thank you for your reply, which makes everything clear to me. Thank you for taking the time to reply really in depth and detailedly.

Indeed I watch the tutorial ( https://www.astropixelprocessor.com/part-4-light-pollution-correction-and-background-calibration-of-the-mosaic-tutorial-milky-way-to-rho-ophiuchi-by-mabula ) some weeks ago. I remember that I learnd that setting the boxes is a iterative process with trial and error in order to get as most green boxes and as less yellow or red boxes as possible. But I must admit, eliminating all non-green boxes is almost impossible since when I delete yellow or red boxes and reprocess some of my former green boxes turn red or yellow.... but that's a different story.

I simply got confused where to set the boxes because I can remember an older video tutrial (by you? or by Sarah Wager?) where it seems that boxes only are placed on light pollutated areas. But obviously I misunderstood that.

Thank for you help,
Stephan

This post was modified 3 months ago by schurig

ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 2 years ago
Posts: 2081
March 14, 2019 13:38  

Hi Stephan @schurig,

You are most welcome 😉

In my tutorial, I don't indicate that your principle goal is to reduce the amount of yellow and red boxes. Yellow boxes and even red boxes can be boxes that are placed correctly, the color however indicates the likelyhood of the box being correct.

So if you see a red box, you sometimes need to double check. so remove it and place a new one, perhaps place it in a slightly different spot with different size for instance.

The colors are based on statistics of a normal (gaussian) distribution, a yellow box means that the area select box is more than 1 standard deviation away in the correction model..

A red box is more than 2 standard deviations away.

Data within 1 standard deviation (sigma) is about 68% of the population and data within 2 sigma is about 95%. See wikipedia about this:

Wikipedia 68-95-99.7 rule

So that means for any correction model:

32% of the boxes, (so 1/3) will still turn yellow normally.

and 5% (1/20) will still turn red 😉 It does not mean that a red box can't be correct !

Mabula

This post was modified 3 months ago 3 times by Mabula Haverkamp - Admin

Main developer of Astro Pixel Processor and owner of Aries Productions


Share: