Registration failures in new beta

Hi,

I am getting a problem with star analysis in beta2. It worked fine in beta1. I sent a bug report today, hopefully it will be fixed in beta3.

@karolbe thanks - for me the star analysis is succeeding but the registration of many light frames is failing.  Sounds like I need to raise a bug report also.

We probably need some of that data to have a look as to why that happens. To my knowledge there weren't many changes that would result in less data to be registered.

@vincent-mod sure - let me know how I can provide.  I've run the same data after rolling back to 1.082 as a check and it throws no errors at registration.

You can upload a small set of subs (including masters calibration frames) that show the failure to our server here;

Go to https://upload.astropixelprocessor.com and use upload1 as username and upload1 as password.

Create a directory named “scott44-registrationfailure” and upload in there. Thank you!

@vincent-mod I'm uploading a handful of lights and some calibration masters.  I reinstalled the beta and this time I got a star analysis failure on one sub (0020 in the dataset).  So the behaviour is not consistent as this morning it passed Star Analysis on all subs but then failed registration on several.  Scractching my head to be honest.  1.082 does not throw any errors with this data which is shot through a Samyang 135mm lens.

Thanks for uploading, we'll test it as well and maybe Mabula can see what is going on as well. Thanks!

I am also seeing a lot of registration errors in this beta, and need to either increase analyse star count or scale stop. Exposures captured in nautical darkness are worse affected, but I was ditching them anyway.

Happy to upload some subs for someone to check but maybe we are all integrating a few subs whe.mn the sky brightness is higher than usual?

I don't have the previous beta release to compare though. 

Yep, @cwm2co1 Colm, you can upload to the above link as well, but instead use a directory named "cwm2co1-registrationfailure".

We'll look at all data to see what is happening and if this is beta related, will take more time then usual as I need to confirm this also with Mabula, stay tight.

Ok, @scott44 , I'm indeed not able to even analyse the stars in the data you uploaded. Not even without calibration frames (in the event those have an issue). I'll pass this on to Mabula for sure.

I've started seeing the same behavior as well.

A few days ago I stacked a session of 61 light frames successfully. Today, the many of the same files fails at registration when I decided to re-stack and reprocess (I accidentally deleted the previous stack). No data has been removed or added. It's the exact same files as I did the first time. I am running it again now, and if it fails again I'll also revert to the previous version as the above poster did.

Weird. I uninstalled beta2 and reinstalled beta1. Stacked like a charm, no errors. I even tried a multiple session of the same target (2 sessions of lights/darks) and the results was just as awesome as I've known APP to produce.

@wheeljack  I think the issue is specific to beta2.  Previous versions are fine.

I'll ping Mabula again about this. Thanks.

Hi all, @scott44, @karolbe, @vincent-mod, @cwm2col, @wheeljack,

All the issues are related to a change in the star analysis module in 1.083-beta2. I implemented the change because it was working on my test data and fixed a problem on other problematic data. As it is, it turns out it is not stable enough clearly for other datasets.

So I will roll-back the star analysis to the 1.083-beta1 version, or 1.082 and then will test all data that we now have to see if all is okay 😉

Thank you all for reporting this !

Mabula

@mabula-admin Thanks Mabula 🙂