Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

PixInsight - PhotometricColorCalibration - White Balance Function Graphs

astrophotography dslr
  • Please log in to reply
5 replies to this topic

#1 Jim Waters

Jim Waters

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2600
  • Joined: 21 Oct 2007
  • Loc: Phoenix, AZ USA

Posted 13 August 2019 - 07:16 PM

These graphs are created at the completion of PhotometricColorCalibration.  I need some help interpreting these graphs.  What's good, what's bad ...etc.  I am getting lost reviewing the on-line documentation.  For example...  Are these graphs OK?

Attached Thumbnails

  • PMCC Output.jpg

Edited by Jim Waters, 13 August 2019 - 07:17 PM.


#2 Brett Waller

Brett Waller

    Mariner 2

  • *****
  • Posts: 232
  • Joined: 18 Aug 2006
  • Loc: Staunton, VA

Posted 13 August 2019 - 07:26 PM

The graphs show the linear transform that alters the color from the values calculated by your software to a best fit match for the colors determined photometrically in the Vizier database.  The line is a linear fit regression and is the actual transfer function used, while the dots are the individual data points (stars) used in the calculation.  There is no "good" or "bad" here, it is just a statistical technique to determine the transfer function to result in the "best fit" of your data to that in the database.

 

I hope that helps.

 

Brett



#3 Jim Waters

Jim Waters

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2600
  • Joined: 21 Oct 2007
  • Loc: Phoenix, AZ USA

Posted 13 August 2019 - 07:37 PM

That somewhat helps.  I already understood that's its best that the 'green dots' fit the line.  I have had some results where the green dots were all over the place.  

 

What about the height and slope of the line?  Is there any good or bad there?  Is there such a thing as an ideal plot?



#4 Brett Waller

Brett Waller

    Mariner 2

  • *****
  • Posts: 232
  • Joined: 18 Aug 2006
  • Loc: Staunton, VA

Posted 13 August 2019 - 08:08 PM

Yes, there would be a line which resulted in no color correction at all, which is the same as saying the "color" on your images already matched the "color" according to the photometric values in the database. 

 

The height and slope are highly dependent on the scales used on both the axes, and I doubt it is always constant, but I rather suspect it is scaled to give the most legible graph.  In general, you aren't striving to get a graph that has a particular slope or height, in the same manner that you aren't looking to get a particular shape on your histogram transformation curve. It is simply a mathematical transform, not a measure of quality.



#5 Jim Waters

Jim Waters

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2600
  • Joined: 21 Oct 2007
  • Loc: Phoenix, AZ USA

Posted 13 August 2019 - 08:15 PM

Thanks Brett.



#6 Brett Waller

Brett Waller

    Mariner 2

  • *****
  • Posts: 232
  • Joined: 18 Aug 2006
  • Loc: Staunton, VA

Posted 13 August 2019 - 08:24 PM

You are welcome Jim. Here's a link that may help:

 

http://pixinsight.com/tutorials/PCC/

 

Where it refers to "second-order and higher terms", it mean that a slight curve might give a better statistical fit than a straight line, but in general, that is nit picking and would scarcely be noticeable.  The method is designed to use a first-order, linear fit.

 

Brett




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: astrophotography, dslr



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics