Batch Preprocessing ImageWindow.windowById(): invalid view identifier.

[pbkwee]

I use the batch preprocessing script.

I create and save a process icon.

Later on I open that icon hit the round button, get the dialog, hit run.

Then I get a ImageWindow.windowById(): invalid view identifier.

I get a writing bias frame message just prior to that error message.

Latest pixinsight.  Batch preprocessing 1.10.

PS: after an error the batch preprocessing option disappears from the script menu.  Seems to reappear after a restart or after re-checking for a software update.

[Juan Conejero]

Hi,

To help you I need more information. The error you're getting is very strange, and might be related to file names. Could you upload a reduced set of images where this problem can be reproduced?

Also, can you put some examples of the *full* names (that is, including all subdirectories) of the files you're trying to use with the script?

[pbkwee]

Problem isn't happening with the latest batch preprocessing.  I see there is a new warning about names having spaces in them.  e.g. 'Filter 2' => Filter_2.  May be related?

[MMys]

Hello,
I noticed exactly the same error also, when trying to calibrate data of my friend with his new 36Mpix ZWO ASI094MC camera.  It occurs at the end of integration, where final image is opened in workspace, but it is not saved due this error. We tried many combinations of BPP settings, and find out so this error disappear, if Generate rejection maps option is unchecked. All other settings have no influence.

Error is probably dependent on size of the images. When I tried to calibrate my smaller 16Mpix images, on the same setup, error disappeared, even if Generate rejection maps was checked.

Error is platform independent, exists on both Win and Linux platform, on all three machines we tried.

BPP is 1.46 on both Linux and Win. 

In the image below it occured after FF integration.  When using master calibration files, it will occur after light integration. It looks like unable to locate resulting window with integrated image in interface, before saving it.

Problem is not related with filenames. We tried to rename files to simple names as FF01 to FFnn, light00 etc...  With no effect 

Also this problem not depends on CFA, nor nothing with registration. We tried handle images as mono, and have selected calibrate only, again without effect.

[MMys]

Images whose suffer by this bug can be downloaded here:

http://astrofotky.cz/tmp/PI-issue.zip

[Konihlav]

Hi there,

I've been using other SW for calibration of my frames, but since I acquired ZWO ASI094MC Pro, a beautiful 36Mpix camera, I had to do some changes (to deal with such amount of large data). First, I bought a new desktop computer (see Benchmark Report), second I started using PI for the calibration process (there are few things that I do not like on some of the scripts as they run only on single core making the whole process 16 to 32 times slower and it's a pain to wait for the results :-( but on the other hand, those multi-threaded processes are superb :-) ! that has actually been the reason I have started to use PI more often than before).

At the link above you can find data and print-screens to demonstrate the issue that happens on both of my machines (Linux + Windows) as well as friends Windows machine. It's well documented there! please have a look (I am a software developer BTW).

The work around is to uncheck the "generate rejection maps".

Otherwise PI is unusable for processing/calibration of data from this particular 36Mpix camera.

[msmythers]

Hi

I thought I would try the most simple of tests. Debayer, StarAlignment and ImageIntergration with rejection maps. No calibrations just simple integration. I have a old Windows 7 computer with 16GB of ram and an AMD 6 Core processor just for information running the latest PI.

Below is the resulting screen capture.


Mike

[Konihlav]

HOW TO REPRODUCE THE ISSUE step by step (for dummies )

0. download the data referenced above (there are three single raw flats, three single raw lights and two masterdarks matching flats and lights)

1. run PixInsight

2. go via Script -> Batch Processing -> BatchPreprocessing

3. load three individual flats, e.g. via Add Flats

4. load three individual lights, e.g. via Add Lights

5. load two master darks, e.g. via Add Darks

if images are needed I will attach them

6. in Lights panel of Batch Preprocessing Script, check Calibrate Only

7. set Global Options as:

check Generate rejection maps (this is reason why it fails!)
check use master dark

set output directory

Go Run

if you uncheck Generate rejection maps then it doesn't fail.

That's as simple as it can be. I attached images and description in the zip file for download above.











:-)

if you can't reproduce it with this description, then you can never do it.

[Konihlav]

BTW my goal is to report a BUG to the developers. My goal is not to find alternative way how to calibrate, debayer and register.

BTW if I am the developer of that BatchPreprocessing script I'd remove the stupid warning message about not using biases. Who the hell cares about biases when just matching darks are enough (and perfectly OK) to go with? Last time I was using biases were with a CCD camera (clean Sony ICX chips). With Kodak/TSI/OnSemi KAF/KAI chips I always needed properly matching darks (scaling works only with some KAF sensors). Also, with current CMOS world, you must always use darks for proper (best) calibration results. Fiddling with biases has nowadays reason only for minority of use-cases and applications.

[msmythers]

I was just pointing out that the script may be the problem not the actual act of generating rejection maps since the script uses these tools and that what I did was just another point of information. All information is good when troubleshooting! I always start with the simplest of situations when trouble shooting and work towards the problem to find where the process goes wrong. That's all I was doing.

I made no mention of this is how you should do your integration. I was just trying to help all those concerned.


Mike

[Konihlav]

Mike: it's all OK what you wrote, I agree to up to almost everything you wrote.

BUT what would make ME HAPPY would be if you said, yes, I took your data and I was able to reproduce the bug in the BatchPreprocessing script... that's something that would count as useful :-) because I do have the workaround, I need other users to report the issue (confirm) so as the developers would fix it (I provided the fully functional test case). Without crowd saying this is a bug it won't be noticed and fixed... thank you!

[sharkmelley]

Out of curiosity, I have a question.  You have selected "No rejection" and then selected "Generate rejection maps".  Isn't that contradictory?  Maybe it is unable to generate rejection maps if no rejection is taking place?

However, I agree it would be nice if it produced a more obvious error message.

Mark