PixInsight 1.8.5: DrizzleIntegration not working anymore!

[kayronjm]

I believe I have discovered a bug in DrizzleIntegration in the newly released 1.8.5.1352. I went through my regular procedure to pre-process data, this time adding in LocalNormalization:

1. ImageCalibration
2. CosmeticCorrection
3. SubframeSelector to add a custom weighting parameter I call SSWEIGHT
4. StarAlignment with Generate Drizzle Data enabled
5. LocalNormalization
6. ImageIntegration with both Drizzle and LocalNormalization data added, rejection algorithm to Average Sigma Clipping, Average and Local normalization selected, FITS keyword selected and set to SSWEIGHT
7. DrizzleIntegration with both Drizzle and LocalNormalization data added, default settings

The following error comes up every single image it tries to integrate:



I have tried removing LocalNormalization completely from the workflow but the error is the same. In fact, what I used to do before updating to 1.8.5.1352 no longer works as DrizzleIntegration always fails with the same error. I used to always deal with FITS files, but am now forced to deal with XISF files, just in case the bug is linked somehow!

[Juan Conejero]

Hi Kayron,

Bug confirmed and solved. I'll release an update ASAP.

However, this can only happen if you:

- Disable pixel rejection in ImageIntegration. The console message "** Warning: The drizzle data file contains no pixel rejection data." tells you that your integration has not been normal.

- Enable local normalization in DrizzleIntegration.

The combination of both factors leads to a segmentation fault. We simply didn't include this case in our tests. If you perform a normal integration, that is with pixel rejection, DrizzleIntegration works perfectly with local normalization enabled.

Anyway, this bug is now fixed, thank you for detecting it.

As for the FITS format, the problem is simply that we cannot implement the new features and tools that we are developing with FITS. Well, of course everything is possible, but in a suboptimal and extremely inelegant way, and generating even more interoperability problems, which are completely unavoidable with FITS. With XISF, everything is clean, efficient, and subject to unambiguous standard rules. If you want to stay in PixInsight, better forget FITS as soon as possible.

[kayronjm]

Thank you for the quick attention Juan! It is much appreciated!
It's odd because I definitely had Pixel Rejection enabled in ImageIntegration, with Average Sigma Clipping set. I tried with and without LocalNormalization. Anyway, what's fixed is fixed!

And no problem about XISF. I have always used FITS but honestly do not care which format is used as I work with PixInsight exclusively regardless.

[Juan Conejero]

The update is now available for all platforms. Actually, the bug was a bit more complicated than what I said, but it was triggered to .xdrz files without pixel rejection data. Thank you for detecting this issue. Even for marginal cases, fixing these problems always makes our toolset more robust.

[kayronjm]

You're very welcome, and honestly, thank you for your quick and robust support. It really does add value to the product!