How to roll back WBBP2

Jim_

Active member
Sep 26, 2012
44
1
I've used WBBP for a long time now.
The new update killed me.
All of a sudden my images are WAY off in color, so much so that I can't color calibrate.
Also now I'm seeing a bias pattern in my images that has never shown up before.
I went back and reprocessed an image set that I had just completed before the update, and it shows the same problems that my new image set does.
I have also had complete system lockups twice now while running WBBP2.
I'm on Win 10, 32g ram. I should not be locking up.

Is there any way to roll back the update?
I have never had a problem with the old ver.

Jim
 

robyx

PixInsight Staff
Jul 17, 2019
159
40
44
Zurich
Hi @Jim_, I'm willing to help you to have a smooth migration to the new version if you would.

It's hard to provide the reason why you're facing these issues without having a bit more information about your session. In particular, if you would share the log files taken from your last runs of both WBPP 1.5.3 and 2.0.2 on the same file set I can compare them and try to find why you have different results.

I am not sure if PI provides a way to roll back scripts, but if you decide not to go with this new version I can upload here the previous version 1.5.3 that you can feature and you can disable the 2.0.2 (both operations could be done from the script menu).
 

Jim_

Active member
Sep 26, 2012
44
1
Hi @robyx, where would I find the specific log files?
There's probably a setting I'm missing somewhere.
This is a Canon 60da camera.
I just ran the same set through DSS and it came out fine.
I did select CFA images in the control panel.
Jim
 

Jim_

Active member
Sep 26, 2012
44
1
So, I went back and did some more investigation.
I'm dealing with three nights of data.
If I process them one session at a time everything comes out fine.
If I process them all three sessions at a time, using the keyword session, that's when the bias pattern and the strange color show up.
I've changed all the file names so I can separate the sessions, using lights and flats.
The bias and dark frames are common from a library.
I've tried with dark optimization both on and off.

Again, I'm sure there's setting I'm probably missing.
Any help is appreciated.

Jim
 

Attachments

ngc1535

PixInsight Ambassador
Feb 1, 2014
751
133
AdamBlockStudios.com
Jim,

You are subtracting a 301 second dark frame from your Flat field images. This is likely not a good idea (I am being kind here).
For flats, best practice is to subtract an exact matching dark frame or the bias. If you want to subtract the Bias, you need to uncheck the use Dark check box. I do not recommend optimizing darks to calibrate flats if you do not have to do it. (Flat cannot have appreciable dark current in 0.07seconds!).

Please remember to click on the Calibration Flow Diagram for each group to make certain it is how you want it (if you find the files that are lit up confusing).

I am reading this wrong? I just want to get this out of the way before going further...
-adam
 

Jim_

Active member
Sep 26, 2012
44
1
Adam,

I've tried both with and without subtracting the dark from the flats. I was in a rush taking the screenshots and I probably posted one that was confusing. I realize the flats are not the exposure length that is optimal but that's not the real problem right now.
I don't understand why I can run the same exact image set by it's self, with the same exact settings, and everything runs fine.
Once I get all three separate image runs finished I can combine them and the image is fine.
It's only when I try to run all three sets at one time that the problem comes up.

Jim
 

ngc1535

PixInsight Ambassador
Feb 1, 2014
751
133
AdamBlockStudios.com
Jim,

First of all, you blew off my comment...but I think it is important. I don't think this is what you want to do and it will lead to problems. In both cases subtracting a 300sec dark from a subsecond flat OR not subtracting anything...in both cases *neither* calibrates the flat.
You show you have a bias frame, if your camera is well behaved why not use it?

But... lets assume you are absolutely correct and it does not matter. Your contention is that when you preprocess a single session... that a specific image in this set is DIFFERENT than the outputted grouped image (the same one) when you try to do it in one go using a keyword to group by session.

So, can you please show that the image is either the identical (as it should be ) or different in the two cases?
You have these two images saved in their corresponding output folders in each case. What do they look like?
You should be able to say that all images in these two different routes are either identical or different.

-adam
 

Jim_

Active member
Sep 26, 2012
44
1
Adam,

I'm not blowing you off at all. I think I may misunderstand what you're saying though.
If you look at the "multi sessions" image I attached earlier, it shows I have 30 bias frames loaded.
I took a screen shot of the flow diagram that shows the bias frame is being subtracted from both the lights and darks.
I may be reading the flow incorrectly though. I don't do this everyday and I rate my level somewhere between a beginner and an intermediate user.
I may know just enough to be dangerous.

I did attach two cropped images above that shows the final images of both.
Should I attach links to the full size images for you to look at?

Jim
 

Attachments

fredvanner

Well-known member
Apr 17, 2019
1,073
136
70
Wells, Somerset, UK
Jim,
I'm finding it really difficult to understand exactly what you are saying. If I understand what you say above:
  • you have three sets of 15 light frames of 301s captured in three separate sessions;
  • you have three sets of (very very short exposure) flats (21 for session 3, 23 for the other two sessions);
  • you have 22 dark frames of 301s;
  • you have 30 bias frames.
If I process them one session at a time everything comes out fine
I presume that you mean you process each session separately. Your "1 group" screenshot (for the session 2 lights) is not using the separate bias and dark frames, it is using a master bias and master dark which have already been generated (by some process you have not detailed). It also uses a master flat, presumably from the session 2 flats, but pre-integrated using an unspecified calibration.
Your "multi-session" screenshot uses the separate, unintegrated, bias, dark and flat frames, with an "unorthodox" (I'm not confident or arrogant enough to just say plain wrong) calibration configuration for the flat frames.
It is no surprise to me that you get different results.
If the "1 group" result looks reasonable, that suggests that the calibrating masters (bias, dark, flat) were constructed correctly.
All you need to do now is specify the correct calibration processing in your "multi-session" set and you should get the same results.

Bottom line: the results are different because the calibration masters are different, and my bet is that it is the flat processing in the "multi-session" set-up that is causing your problems.
 

Jim_

Active member
Sep 26, 2012
44
1
I was in a rush when first posting the screenshots so I did miss that master frames were being used.
I reran the image set just now with the same settings as the multi session one. No master frames are involved.
The results are the same though. Running the one image set at a time produces an image the is correct in both color and not being able to see the bias pattern.
I'm sorry if I'm not explaining this very well.
I just put the images in the correct area and let the software run. The lights go in the lights, the flats, go in the flats, etc.

The images are the results of one set being ran with no masters involved.
One is the work flow, and the other is a screenshot for the image right after WBPP2 finished with stf only applied.
 

Attachments

ngc1535

PixInsight Ambassador
Feb 1, 2014
751
133
AdamBlockStudios.com
Jim,

Can you make your data available?

When you run one image set at a time... you will have a calibrated and debayered image called "seagull xxx.xxx.fit" or whatever.
When you run it as a mult-session this same image ("seagull xxx.xxx.fit" ) is in Session 2 and it is calibrated and debayered.
Does this image look different?

Also, when you ran the multi-night way you will end up with three master light images. What do the three integrated images for the multi-night output look like?

-adam
 

ngc1535

PixInsight Ambassador
Feb 1, 2014
751
133
AdamBlockStudios.com
Adam,


I took a screen shot of the flow diagram that shows the bias frame is being subtracted from both the lights and darks.
I may be reading the flow incorrectly though. I don't do this everyday and I rate my level somewhere between a beginner and an intermediate user.
I may know just enough to be dangerous.


Jim
Jim,

You are only showing the flow diagram for the Lights. Click on a Flats group and then display the diagram. You will find you are calibrating the Flats with a long master dark and not the bias as you intend.

-adam
 

Attachments

Jim_

Active member
Sep 26, 2012
44
1
Here's a link to all the files.
This is all raw data so the file is pretty large.

Jim
 

Jim_

Active member
Sep 26, 2012
44
1
It's larger than the free dropbox account will allow.
I got a "your dropbox is full" message.
Let me see if I can find another way to share it.
The file is over 3gig.
 

ngc1535

PixInsight Ambassador
Feb 1, 2014
751
133
AdamBlockStudios.com
Everything worked as expected.
There is no difference between running all three at once using keywords... or doing each of them one at a time.
(the top two images in the first screen shot are the same data, but run through the two "different" routes)

Note the setup for the calibration of the flat.
 

Attachments