2023-01-19: APP 2.0.0-beta13 has been released !
!!! Big performance increase due to optimizations in integration !!!
and upgraded development platform to GraalVM 22.3 based on openJDK19
We are very close now to releasing APP 2.0.0 stable with a complete printable manual...
I'm not sure the right way to file a bug - please let me know if there is a better way, or what other diagnostic procedures I could carry out that would help.
I got an ArrayIndexoutOfBoundsException in MultipleViewFullCameraCalibrationSimpleDistortionCorrectionNotSameCameraIncomplete while trying to integrate R,G,B, Ha, and OIII data. The RGB files are from OSC images one night, the Ha and OIII data from a multi-band filter on a second night, and the camera was not exactly aligned the same the two nights.
Mm, not the usual error we see when combining RGB with narrowband. But just in case; you can't normalize narrowband and RGB together and we need to make this way more obvious (it's on Mabula's To-Do list). What you should do is to split the RGB into their separate channels to make them mono (tab 2, split channels option) and then reloading (after a clear or restart of APP) them into their channels. These can then be processed together.
@vincent-mod Thanks - but that's what I did. The R, G, and B files are split from RGB files from my Fuji camera, then the Ha and OIII were split (using tab 0 algorithms) from multi-band filtered images (an L-eNhance filter). The two sets of images were not taken on the same night and are not aligned, so the "Combine RGB" tool wouldn't work on them. So, I was trying to align all of them together so I could crop out a piece that had all five images (another vote for having APP calculate the intersection of the images, not just the union). I'm trying to register all the files to the same reference image, then figure out the intersection, mass crop, then combine-RGB. Suggestions for a better workflow would be appreciated!
Ah ok, so is the overlap that different between the sessions? A regular workflow would be to load all these into their channels, go to tab 6 and press integrate. Maybe change a few parameters but that's about it. If the field of view is so different, then it would almost be a mosaic kind of workflow. Do you platesolve when targeting an object?
@vincent-mod That's what I was trying when I got the error. I'm going to try integrating each channel separately (while using the same reference frame), then crop and merge. We'll see if that helps.
That might also work, but weird it doesn't. If you want to upload the frames I can have a look as well.
Vincent - I'd love it if you can look at the frames. Can I send you a dropbox link or similar?
I've got a subset that fails every time:
one Halpha frame + flat that I want to use for reference, five blue frames (extracted from Fuji RGB files) + flat, a bias and a bad pixel map. I load the Halpha and blue lights, and all the calibration files. Analyze goes fine and I select the Halpha frame for a reference in (4) Register. That runs fine, then it dies in Normalize (see screen shot). In Register I have Dynamic Distortion Correction On, Same Camera and Optics Off, Registration Mode = Normal, Registration Model = Calibrated Projection.
Maybe the issue is the RGB lights extracted from the original lights in the Calibrate step? I processed the Halpha and OIII lights without problems. Any help would be appreciated! This has been *very* frustrating!
@wvreeven Thanks! The files are uploaded. One Halpha frame + flat that I want to use for reference, five blue frames (extracted from Fuji RGB files) + flat, a bias and a bad pixel map.
Thanks Norris, let me download and check them out..
edit; Oh, your folder on the server is empty, are the file in the main directory outside of the folders from you by any chance?
@vincent-mod Sorry - so many newbie mistakes! I've re-uploaded the files to the right directory, but can't move or delete them from the top-level. The highlighted files in the screenshot are the ones that need zapping. Thanks!
No worries, thanks for pointing those out. 🙂 I'll remove the top ones and will check your data asap.
@vincent-mod I *am* able to make this work with a different set of files: I shot both RGB (with a light pollution filter) and Ha-OIII in one session last night, just swapping filters. I tried both integrating one channel at a time, or all 160+ subs at once to five different integrated files (R,G,B,Ha,OIII), and the RGB Combine tool works fine. I used Reference as the composition mode, so all the files were the same size, and maybe that was the difference. I'll redo the set I uploaded and see if this fixed things.
Edit: No, still fails with the set I uploaded, again in the Normalize section before the composition mode setting is even used. I don't see what's so different that one set throws an array index error and the other finishes just fine.
Sorry for not coming back to you sooner; I'm looking at it now and what strikes me is that the lights you uploaded were already calibrated. So I normalized them without calibration frames. That works with just the blue lights, but adding the H-Alpha indeed introduces an error.
Looking at that frame a little closer and I see the issue, APP thinks it's an RGB image and you can't normalize mono with RGB at the same time. Interestingly, the FITS header shows this should not be a CFA image. So I reported this to Mabula. In the mean time I reloaded the h-alpha an selected "no algorithm" in tab 0, for the algorithm and that works. I'll let you know if Mabula is able to see why.
@vincent-mod Thanks, Vincent - I appreciate your detective work. I'll try running the files again without double-calibrating the RGB split files, and no algorithm for the Ha and OIII files. But I wonder why this set triggered the problem, but the second set (all shot on one night) did not.
Thanks again, Norris
This continues to be very frustrating, but I did get some small success. I'm puzzled by when the tab (0) algorithm setting is acting and what it is doing - that's a part of my issues.
For Ha and OIII, used tab(0) Extract Ha and OIII, added calibration files, and in tab(2) chose "split files" -- this resulted in 16b Gray files which were not CFA.
For R, G, and B, used tab(0) Adaptive Airy Disk, added calibration files, and split the files.
Cleared everything, then loaded Ha, OIII, R, G, and B split files, no calibration files, and set tab(0) algorithm to Adaptive Airy Disk. Had to integrate each set of files separately, since more java out-of-bounds errors were thrown. Problem is, that sets were not aligned to each other.
Cleared everything, loaded separate Ha, etc. integration files as lights, re-integrated each channel - now I had five aligned integrated channels and the Combine RGB tool was finally happy. The image needs more tweaking, but at least it didn't crash.
Edit: Lots of manual hand-holding here, but this seems pretty robust. I've combined Ha, OIII, and RGB data from three nights together without any crashes.
Excellent. Yes, the ways in which errors pop up aren't the most user-friendly I'd say, but we are going to make the process clearer as many run into this. Good to see you have a good workflow now at least!