'Unknown exception' : CometAlignment refuses to "Add files" [SOLVED -SEE POST#4]

[fkieffer]

Hi,

so I just finished preprocessing some pictures of 21p Giacobini Zinner, everything works fine (on Macbook Pro, PI version 1.8.5.1352), until i ask for some "CometAlignment"

I can't add files : when i'm prompted to select the files, i get an "unknown exception" error message, and I don't know what's causing this !!
Console says :

Code: [Select]

Loading image: w=3906 h=2602 n=3 RGB Float32
63 FITS keyword(s) extracted.
So to be clear, here's how it comes :
clicking on "Add files" -> selecting the files -> clicking "ok" -> console output (see above) -> immediate error message -> 

I can actually select a list of files with any other process, only CometAlignment is causing trouble...
I tried
- rebooting
- reinstalling
- starting PI from another session (user)
- changing the file format, reverting to fit instead of xisf

Nothing seems to have any effect. Ideas ? Thanks for reading (and thanks+++++ for answering  )

[pfile]

what happens if you select only one file? is there a # of files beyond which this happens?

rob

[fkieffer]

Thanks for your answer rob !
Well, then, the result is exactly the same...
I tried selecting only a few, then only one : my guess is, the process must start "reading" some info (headers ?) in the files before adding them in the list (otherwise the console would not show the frame dimensions...), and then continue with the second file and so on, but in my case, the process crashes right after the first file.

Here is a link to the files.
https://mega.nz/#F!fkRBhYAQ!U-hLih8hoHa-IhFBDOKGqw

There must be something wrong with my files. I just tried again a few minutes ago : i can actually add the raw files from the sequence (canon EOS 1000D, .CR2).
I cannot add the calibrated .xisf files, nor any version of the same frames once they have been preprocessed...
Neither can I import any other .xisf or files from any other project...

[fkieffer]

THAT'S IT ! I think i found a solution.   8) 8) 8) 8) 8)

I could not understand why my xisf files were not accepted, whereas a raw photo exported to xisf would fit without a problem.
The raw files were not converted to xisf before applying the calibration process.

So, using "batch format conversion", i converted all my "Light" pictures to the .xisf file format, and then only go on with all the processing stuff. And this seems to work now.

But what is the difference between a calibrated .xisf image that already was .xisf, and another that wasn't ?

[pfile]

strange - ImageCalibration should be able to read in CR2 files and write out XISF files without the need to have the CR2s converted to XISF. i do know that CR2 files can not be read incrementally the way fits and xisf files can, so there is a slight difference in how they are treated, but usually this only causes problems when you are say, integrating CR2 files directly and the memory footprint blows up.

so you are saying that you did CR2->XISF, then imagecalibration on the XSIFs, then registration, etc?

rob

[fkieffer]

yup, that's it, and only by doing so, i would be able to perform a CometAlignment

but now, some other strange thing happened : Linear clip fitting will only work with 110 of my 111 frames... #93 causes another "unknown exception" that's weird as well ! (took me some time to figure it out, but i'm getting suspicious : i have never had any bug before, and now I just had 2 of them, in less than a week…)

[pfile]

is it a new machine? have you ever run prime95 or memtest86?

rob

[fkieffer]

Nope, the machine isn't quite new, it's a Macbook Pro bit less than 2 yrs old, never had any issue with PI.
But i had never tried to take pictures of comets before...

... well, at least i know how to make it work !

[bulrichl]

The two xisf files (I refer to the one that works as 'File_1' and the one that doesn't as 'File_2') differ only in the XISF Header, namely:

--------------------------------------------------------------------------------------------------------------------
File_1:

<FITSKeyword name="DATE-OBS" value="'2018-08-12T22:28:17'" comment="Camera timestamp"/>
<FITSKeyword name="HISTORY" value="" comment="Calibration with ImageCalibration module version 01.04.01.0337"/>
<FITSKeyword name="HISTORY" value="" comment="Noise evaluation with ImageCalibration module version 01.04.01.0337"/>

--------------------------------------------------------------------------------------------------------------------
File_2:

<FITSKeyword name="DATE-OBS" value="'2018-08-12T22:28:17Z'" comment="Camera timestamp"/>
<FITSKeyword name="HISTORY" value="" comment="Calibration with ImageCalibration module version 01.04.01.0332"/>
<FITSKeyword name="HISTORY" value="" comment="Noise evaluation with ImageCalibration module version 01.04.01.0332"/>

<Property id="Instrument:Camera:ISOSpeed" type="Int32" value="1600"/>
<Property id="Instrument:Camera:Name" type="String">Canon EOS 1000D</Property>
<Property id="Instrument:ExposureTime" type="Float32" value="120"/>
<Property id="Observation:Time:Start" type="TimePoint" value="2018-08-12T22:28:17.000Z"/>
<Property id="PCL:CFASourcePattern" type="String">RGGB</Property>

--------------------------------------------------------------------------------------------------------------------

I showed only the differing rows.

In file_1, the 'Z' is missing in DATE-OBS.
In file_1, ImageCalibrationModule version is 01.0401.0337 instead of 01.0401.0332 in file_2.
The last section of File 2 is missing in file_1.

No other differences found, that means: there are no differences in the image data.

The culprit is the 'Z' in DATE-OBS, for when I try to add file_2 in CometAlignment, the following error appears:

"Can't convert DATE-OBS keyword value '2018-08-12T22:28:17Z' to a Julian day number."


Bernd

[fkieffer]

thank you bulrichl ! This makes perfect sense with when / how the error occured.