ImageIntegration should consider the path to assign drizzle files to fits

[Ioannis Ioannou]

I have two sets of files, from different nights, in different directories. They overlap regrading names, eg M13-01_R exists in both. I want to integrate from both nights. But ImageIntegration does not look in the path when it is trying to assign the drizzle files to the fits. As an example, I have added in ImageIntegration two M13-01_R_cc_r.fit files, from path1 and path2. I have added also M13-01_R_cc_r.drz from path1. But If I'll add M13-01_R_cc_r.drz from path2, it will assign it to path1's fit, overwriting the previous assignment and leaving the fit from path2 unassigned.

[MikeOates]

A way round this would be to rename the files before calibration, i.e.

"M13-01_R.fit" to "M13_date1-01_R.fit"
and for the other set "M13_date2-01_R.fit"

Mike

[Ioannis Ioannou]

Yes Mike, I know, I have already renamed the second set in order to proceed.
Nevertheless, the bug, although minor, is valid.

[Juan Conejero]

Hi Ioannis,

The new version of PixInsight comes with an update to the ImageIntegration tool that solves this problem. The solution is a new option on the ImageIntegration interface that allows you to use full file paths stored in .drz files instead of just file names. However, keep in mind that by using this new feature your data set will be tied to specific locations on the local filesystem. By using just file names, you can move your images freely to different locations without problems.

So as a general rule, I strongly recommend you store all of your raw frames and images using unique file names. This allows you to apply more flexible processing pipelines.

Thank you for pointing out this limitation.

[bitli]

Some self promotion: Look at the script FitsFileManager to rename the image files systematically (while keeping the original name in a FITS keyword for easier cross reference if needed).
-- bitli

[Ioannis Ioannou]

Thank you guys. Actually I have used a quick shell for loop command to rename the second set, Unix shell is always my friend .

I just wanted to point out the limitation for Juan and for anyone else that may missed it.

[Thorsten Lockert]

Hi Ioannis,

The new version of PixInsight comes with an update to the ImageIntegration tool that solves this problem. The solution is a new option on the ImageIntegration interface that allows you to use full file paths stored in .drz files instead of just file names. However, keep in mind that by using this new feature your data set will be tied to specific locations on the local filesystem. By using just file names, you can move your images freely to different locations without problems.

So as a general rule, I strongly recommend you store all of your raw frames and images using unique file names. This allows you to apply more flexible processing pipelines.

Thank you for pointing out this limitation.

It is worth noting that this has come up before, in http://pixinsight.com/forum/index.php?topic=5030.0 -- nothing to do with Drizzle. I don't know if the fix you just provided addresses this case...?

Thorsten

[Juan Conejero]

I don't know if the fix you just provided addresses this case...?

No it doesn't. For some reason I overlooked that thread, and this problem has never arisen in our tests because we never have images with duplicate file names. I'll try to solve this problem with BPP as soon as possible.

That doesn't change the fact that having duplicate file names is a source of potential problems. All raw frames and images should always have unique file names. Along with the solution that Ron describes in the original thread, you can use the FitsFileManager script as bitli has pointed out.

[pfile]

for me it was something that happened when using lots of "stock" equipment. for instance the canon sequencer that comes with their digital cameras starts over every night with IMG_0001.CR2. maybe it's configurable, IDK. but as you say, when you use some 'proper' astro capture software that puts dates or does not reset sequence #s, then you're ok.

rob

[Thorsten Lockert]

That doesn't change the fact that having duplicate file names is a source of potential problems. All raw frames and images should always have unique file names. Along with the solution that Ron describes in the original thread, you can use the FitsFileManager script as bitli has pointed out.

Yes, I am aware of those, and have used both. I really only encountered this issue when I was in the process of setting up the observatory, before I had configured "everything".

Thorsten