Batch PreProcessing error

[papaf]

Actually,  now that you mention this, I too have a data set that doesn't align properly, even if, during alignment, pairs of stars are found, albeit little. It was an Ha test captured during full moon, so I just assumed it was bad data. Maybe it's something related to the newly introduced star number filter?

[Juan Conejero]

"File I/O error: Invalid or empty file name"

This bug has been fixed in version 1.31 of the BatchPreprocessing script, which I have just released as an update.

In addition, by default the script now uses triangle similarity for star matching during image registration, so it supports mirrored images. Finally, the script provides access to the new noise reduction parameter of StarAlignment, which you can use to align images of very low-SNR. You'll find these parameters on the Lights page, in the Registration parameters panel.

Hope this helps.

[georg.viehoever]

Hi Juan,
I believe it would not be so difficult to add mirroring support for non-triangle matching as well. Just try to match programmatically mirrored versions of the image. This would at most double the runtime.
Georg

[cfranks]

hi,

no, Scott, you are not the only one running into problems using the preprocessing script and CR2 files.
Have 1.8 RC7 and PP script 1.30

I sometimes get the File I/O error message, and also "cannot integrate less than three files" while there are 67 selected.
Sometimes I can get through by just clearing all the files from the list, and simply re-selecting them.
Then running the script again without changing anything else.
Fortunately that happens not too often, and then the script works like a charm.
Unfortunately I am not able to specify what could be a cause.

Just noted: File I/O error occurs after registration, so the c_d_r.fit files produced by the PP script may be used with the integration module,
and that leads to a successfully integrated master light. The PP script just could not do the integration for some reason.
My filenames are all as given by the Eos ;-)

cs, Tommy

Tommy, 

In my experience, your 'cannot integrate less than 3 files' message is because you are using a Master Dark, Bias and/or Flat and have not ticked the appropriate 'Use Master ....' box.

Charles

[jerryyyyy]

I just loaded 1.31 and have intergrated some files OK, but these data from the French PixInsight people I could not.  This is the error message.  I used the same reference file and could integrate the Green images.  Too few matches??  Settings?

StarAlignment: Global context
Loading reference image:
C:/Users/yesavage/Desktop/Pixinsight.France/2013/PI Output/calibrated/light/NGC1333_Green01_c.fit
Reading FITS: 32-bit floating point, 1 channel(s), 2048x2048 pixels: done
Structure map: done
Detecting stars: done
284 stars found.

Registering target image 1 of 3
Loading target file:
C:/Users/yesavage/Desktop/Pixinsight.France/2013/PI Output/calibrated/light/NGC1333_Blue02_c.fit
Reading FITS: 32-bit floating point, 1 channel(s), 2048x2048 pixels: done
Structure map: done
Detecting stars: done
282 stars found.
Matching stars: done
61 putative star pair matches.
Performing RANSAC: done
*** Error: RANSAC: Unable to find a valid set of star pair matches.

*** Applying error policy: Continue on error.

Registering target image 2 of 3
Loading target file:
C:/Users/yesavage/Desktop/Pixinsight.France/2013/PI Output/calibrated/light/NGC1333_Blue01_c.fit
Reading FITS: 32-bit floating point, 1 channel(s), 2048x2048 pixels: done
Structure map: done
Detecting stars: done
291 stars found.
Matching stars: done
71 putative star pair matches.
Performing RANSAC: done
*** Error: RANSAC: Unable to find a valid set of star pair matches.

*** Applying error policy: Continue on error.

Registering target image 3 of 3
Loading target file:
C:/Users/yesavage/Desktop/Pixinsight.France/2013/PI Output/calibrated/light/NGC1333_Blue03_c.fit
Reading FITS: 32-bit floating point, 1 channel(s), 2048x2048 pixels: done
Structure map: done
Detecting stars: done
282 stars found.
Matching stars: done
58 putative star pair matches.
Performing RANSAC: done
*** Error: RANSAC: Unable to find a valid set of star pair matches.

*** Applying error policy: Continue on error.

===== StarAlignment: 0 succeeded, 3 failed, 0 aborted, 0 skipped =====
13.17 s

*********************************************************************
* End registration of light frames
*********************************************************************

[tommy_nawratil]

hi,

muchas gracias for addressing the problem :-)

I've now installed the v1.31 of PP script and successfully run several DSLR sets. I will try the older set that caused problems, hope the problems are gone.
Thanks for the hint regarding master dark ticked, but they were 20 nice DSLR darks.
NR in alignment is surely a good addition!

Tommy

[tommy_nawratil]

hi,

have now processed several sets of DSLR pics, no more problems! 

here are some proposals for future implementation:
- drawing the triangle to the workspace, and reopening script later doesn't provide selected pics,
only reference and path. Would be nice not to have te need to reload all the lights/flats/darks/bias.

- script blocks PI when opened. Not possible to use blink to select good pics, except opening other instance of PI.

- flat stacking is filter sensitive. Would be nice to have a switch to turn that off,
in case one wants to use flats with no or different filter set in fits keywords.
btw, while we have a script for fits keywords, it is not possible to edit these in PI???

just some small ideas...

saludos, Tommy

[tommy_nawratil]

...and more wishes for the list:

- if CFA is selected and grayscale fits are loaded, the script could simply output:
"Blockhead, which camera was on this time, huh?"
instead of silently debayering my monochrome Sulfur frames!

- no overwriting of files without warning
they just bear the name of the filter, and I use "empty" for all my narrowbands

cheers, Tommy

[pfile]

how would the script (really, the fits reader) know the difference between a cfa fits and a truly monochrome fits? both are monochrome files.

[tommy_nawratil]

good question!

But just regarding CFA fits, not DSLR RAWs at least:

If there is no fits header detected, and the file extension is not fits but CR2, it could be marked automatically being a CFA file?
And if there is a fits header, and the file extension is not CR2 but fits, the option could be left like it was - or even better:
Maybe there is a possibility to detect a bayer pattern - at least Registax alarms me if I load a color video "Bayer pattern detected, process in color?".
That should not be too complicated, I guess. And overrunning should be an option, just in case.

At the same time it would be a nice goodie, if a DSLR_RAW CFA file is detected, the DSLR_RAW preferences could be overrun by the script.
Sometimes one forgets to change them accordingly before running the script, and debayering takes place before appyling the calibration pics.
I am very sure that happens to a lot of beginners! You check your CR2's using VNG, then start the script and forget to do that.
Once the script runs, you cannot check anything because it blocks the surface.

And vice versa of course. I watched my Sulphur narrowband fits files being debayered and RGB pic output, because I left
the CFA checkbox marked... eeeek!

cheers, Tommy