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

WBPP and bad frames

  • Please log in to reply
9 replies to this topic

#1 EEBA

EEBA

    Ranger 4

  • -----
  • topic starter
  • Posts: 340
  • Joined: 25 Feb 2020

Posted 28 February 2021 - 02:12 AM

Hi everyone, I have another question now. I have been using Pixinsight for a few months and now I am paying more attention to each program that I use in Pixinight.
I am still using the old WBPP and don’t understand who it works. I am able to get my image out in the linear form ready to post process. I know it uses eccentricity , SNR , and FMHW to weight images. But at the end of the WBPP when you get your final image. Does WBPP eliminate some frames? Where do those bad frames go to?

#2 endless-sky

endless-sky

    Apollo

  • -----
  • Posts: 1,010
  • Joined: 24 May 2020
  • Loc: Padova, Italy

Posted 28 February 2021 - 03:03 AM

I use WBPP only for pre-processing, up to debayering.

 

Subframe weighting, star registration and image integration I still do manually.

 

For subframe weighting, I use FWHM, eccentricity and star count as metrics. I throw away the obvious bad frames (clouds, star trails, etc.).

 

For the weighting, I use this formula (taken from SubframeSelector Weighting Expressions_v2, by David Ault), but I modified it with Stars (the number of stars in the frame) instead of SNRWeight:

 

16*(1-(FWHM-FWHM_MIN)/A) + 4*(1-(Eccentricity-Eccentricity_MIN)/B) + 20*(Stars-Stars_MIN)/C + 60

 

where

 

A = FWHM_MAX - FWHM_MIN

B = Eccentricity_MAX - Eccentricity_MIN

C = Stars_MAX - Stars_MIN

 

The max and the min of all those values can be found by running SubframeSelector with "Measure" and then by sorting the results by FWHM, Eccentricity and Stars, respectively (in ascending or descending order).

 

I don't use SNRWeight anymore as a measure, because it treats light pollution as a signal and gives a higher grade to more light polluted frames. Star count instead, I found is a good metric because the darker the sky gets (as the object climbs away from the horizon, for example), the more dimmer stars get captured (as they are not washed away by the light pollution closer to the horizon).



#3 EEBA

EEBA

    Ranger 4

  • -----
  • topic starter
  • Posts: 340
  • Joined: 25 Feb 2020

Posted 28 February 2021 - 03:40 AM

I use WBPP only for pre-processing, up to debayering.

Subframe weighting, star registration and image integration I still do manually.

For subframe weighting, I use FWHM, eccentricity and star count as metrics. I throw away the obvious bad frames (clouds, star trails, etc.).

For the weighting, I use this formula (taken from SubframeSelector Weighting Expressions_v2, by David Ault), but I modified it with Stars (the number of stars in the frame) instead of SNRWeight:

16*(1-(FWHM-FWHM_MIN)/A) + 4*(1-(Eccentricity-Eccentricity_MIN)/B) + 20*(Stars-Stars_MIN)/C + 60

where

A = FWHM_MAX - FWHM_MIN
B = Eccentricity_MAX - Eccentricity_MIN
C = Stars_MAX - Stars_MIN

The max and the min of all those values can be found by running SubframeSelector with "Measure" and then by sorting the results by FWHM, Eccentricity and Stars, respectively (in ascending or descending order).

I don't use SNRWeight anymore as a measure, because it treats light pollution as a signal and gives a higher grade to more light polluted frames. Star count instead, I found is a good metric because the darker the sky gets (as the object climbs away from the horizon, for example), the more dimmer stars get captured (as they are not washed away by the light pollution closer to the horizon).


Thanks for replying, but I was trying to know what happens with the frames in WBPP. “ does it trash and remove the bad frames after weighing them and keeps only the best ones?

From your answer, I understand you only use WBPP for calibration. Then you find out the weights in each frame. And manually you remove one by one of the poorest frames ( probably using blink) . Then you debater, star aligne and integrate.
Now I am confused, what is the purpose of the script if you do everything manually?
Please try to address my firs questions as well
Thanks

#4 endless-sky

endless-sky

    Apollo

  • -----
  • Posts: 1,010
  • Joined: 24 May 2020
  • Loc: Padova, Italy

Posted 28 February 2021 - 04:52 AM

In the script there doesn't seem to be a rejecting component to the weighting routine. I think it just grades the images and gives them a weight based on the parameters you choose in the script (FWHM, Eccentricity, SNR and Pedestal). Then you'll have a weight associated to each image (typical values I have seen are 60-65% for the lower end and 90-95% for the upper end).

 

The rejection I think you will have to do it manually, either before you feed the subs to the script (using Blink, for example) or after (before you integrate the resulting frames from all the steps done by WBPP).

 

I do everything manually, because I use the script only to take care of the boring, repetitive parts: calibration, cosmetic correction and debayering. The new version of the script is exceptional as it also allows for multiple night sessions, with matching flats and darks, according to the keywords you associate to the subs.

 

I tend to do the weighting myself, as this allows me more control on the parameters I choose and also to see if there are some really bad off frames. Then I typically blink the rest. After that I do star registration, again manually because sometimes the script cannot find enough stars and doesn't register all the frames. At that point, I finish with image integration, which I also do manually because of all the previous steps I have to do outside the script.

 

To recap, I don't think the script rejects or delete any image from the weighting process. It keeps all of them and just goes through with registration and integration, using the values that it found during the weighting process.


  • EEBA likes this

#5 EEBA

EEBA

    Ranger 4

  • -----
  • topic starter
  • Posts: 340
  • Joined: 25 Feb 2020

Posted 28 February 2021 - 06:15 AM

Thank you, I will try to process same image (M106). Using your technique to see if I see any improvement after end of integration. I noticed Adam Block in his Pixinsight Fumdamentals works the same way. Only used WBPP to calibrate then he goes manually. I assume that’s the style of experienced Astrophotographers.
Again, I will try to do it and get back to comment . Maybe tomorrow after work. (:

#6 jonnybravo0311

jonnybravo0311

    Apollo

  • *****
  • Posts: 1,418
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 28 February 2021 - 12:19 PM

<snip>

I do everything manually, because I use the script only to take care of the boring, repetitive parts: calibration, cosmetic correction and debayering. 

<snip>

How are you setting up your cosmetic correction? Are you doing it on raw lights, or are you doing it on calibrated lights? My workflow is:

  1. Integrate Bias (or use super bias if I'm feeling lazy)
  2. Integrate Dark (no calibration)
  3. Calibrate flat with master bias from step 1
  4. Calibrate light with master dark and master flat
  5. Cosmetic correction

My understanding of how the WBPP script works is that you need to first run the cosmetic correction process on a light frame, then drop an icon onto the desktop that the WBPP script uses. However, don't you want to run the cosmetic correction process on an already-calibrated light?

 

My apologies to EEBA for taking your thread on this slight tangent :)



#7 EEBA

EEBA

    Ranger 4

  • -----
  • topic starter
  • Posts: 340
  • Joined: 25 Feb 2020

Posted 28 February 2021 - 12:31 PM

I got home after work and I started testing WBPP.
I used WBPP without checking only calibration. I checked the weighting box but not the one that says force weight. It saved the new calibrated files. I tried to see how the weights were recorded in each frame. The only place I found weighing numbers was in the logs files. This doesn’t makes sense to evaluate each weighing frame from the log folder. Now I am trying with box force weigh checked up. How I am suppose to select my best frames?

#8 endless-sky

endless-sky

    Apollo

  • -----
  • Posts: 1,010
  • Joined: 24 May 2020
  • Loc: Padova, Italy

Posted 28 February 2021 - 12:43 PM

How are you setting up your cosmetic correction? Are you doing it on raw lights, or are you doing it on calibrated lights? My workflow is:

  1. Integrate Bias (or use super bias if I'm feeling lazy)
  2. Integrate Dark (no calibration)
  3. Calibrate flat with master bias from step 1
  4. Calibrate light with master dark and master flat
  5. Cosmetic correction

My understanding of how the WBPP script works is that you need to first run the cosmetic correction process on a light frame, then drop an icon onto the desktop that the WBPP script uses. However, don't you want to run the cosmetic correction process on an already-calibrated light?

 

My apologies to EEBA for taking your thread on this slight tangent smile.gif

Before running WBPP, I open CosmeticCorrection, I check CFA (since I am using OSC data from a DSLR), I check "Use Auto detect" and select both Hot Sigma and Cold Sigma with default numbers (3.0). I then drag the process to the workspace with the triangle. When I open WBPP, I click Apply under Cosmetic Correction and I select the process from the drop down menu.

 

WBPP then applies CosmeticCorrection automatically, after the calibration processes and before Debayering.

 

I got home after work and I started testing WBPP.
I used WBPP without checking only calibration. I checked the weighting box but not the one that says force weight. It saved the new calibrated files. I tried to see how the weights were recorded in each frame. The only place I found weighing numbers was in the logs files. This doesn’t makes sense to evaluate each weighing frame from the log folder. Now I am trying with box force weigh checked up. How I am suppose to select my best frames?

The weights are stored in the FITS Headers of the files. SubframeSelector uses SSWEIGHT (default). When you integrate the images, you select FITS keyword under "Weights" and type SSWEIGHT. The integration process will then read the value given to each image and treat them accordingly.

 

The keyword WBPP should use is WBPPWGHT, if I remember correctly.

 

I don't know if there is a quick way of seeing the numbers of all your frames. If you load the frames in the SubframeSelector process, you can see all the numbers (FWHM, Eccentricity, SNR, Stars, etc.), but to see the weights you'll have to run the process and give it a formula (such as the one I wrote in my first post) to actual grade the images. I don't know of a quick way to see the numbers given by another process, unfortunately.


  • jonnybravo0311 likes this

#9 ngc1535

ngc1535

    Vendor - Caelum Observatory

  • -----
  • Vendors
  • Posts: 97
  • Joined: 18 Sep 2008

Posted 28 February 2021 - 05:36 PM

 

 

I don't know if there is a quick way of seeing the numbers of all your frames. If you load the frames in the SubframeSelector process, you can see all the numbers (FWHM, Eccentricity, SNR, Stars, etc.), but to see the weights you'll have to run the process and give it a formula (such as the one I wrote in my first post) to actual grade the images. I don't know of a quick way to see the numbers given by another process, unfortunately.

There is a script called FITsKeys that will allow you to the see values of weight or anyother keyword in the header across all images.

 

-Adam Block

Attached Thumbnails

  • Capture.JPG


#10 EEBA

EEBA

    Ranger 4

  • -----
  • topic starter
  • Posts: 340
  • Joined: 25 Feb 2020

Posted 28 February 2021 - 08:37 PM

Thank you endless sky and Adam, I have to go to work now, but I will watch your videos in Fundamentals one more time. They have a lot Great information. Thanks for taking your time to answer questions in the forum(:
Enrique


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






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics