Failing to Detect Stars

I'm having similar problems after upgrading to 1.067. Very seldom had any rejection.

I have another with Star Analysis 

Module will not run with any adjustment to kappa. It will not run at my workstation, but will  run using Remote Desktop from my notebook.

I thought maybe screen resolution or video drivers. Tried different workstation resolution settings with no luck. 

So you're having the same rejection problems as me, AND another issue?

Yes the Analyse Stars issue started before the upgrade but would run with kappa set to 40, now will not run. 

As for your referred problem, I'm getting about 10-15% rejection, which I didn't get before 

That's interesting as well, could you try with the latest 1.068? If it still occurs Mabula might have to have a closer look at it.

I'm not sure which problem you were addressing:  The module not running or the failure to detect stars.

I'm on 1.068 and yes it's still failing to detect stars.  As mentioned previously, for me it could be the data. 

Clear skies expected tonight, I'll attempt the same target and this time I'll collimate the scope, but keep all the other parameters the same.

Posted by: Cheetah

I'm not sure which problem you were addressing:  The module not running or the failure to detect stars.

I'm on 1.068 and yes it's still failing to detect stars.  As mentioned previously, for me it could be the data. 

Clear skies expected tonight, I'll attempt the same target and this time I'll collimate the scope, but keep all the other parameters the same.

I was addressing both as I didn't know the version you were running on. If it's occurring for more people, @bula-admin will very likely have a close look at it.

Vincent,

My original issue was the Analyse Stars. It will not start and locks the app. But if I run it using Remote Desktop from my notebook it runs fine.

I upgraded to 1.067 and that's when I started to get star detection issues. 

The Analyse Stars is still an issue after the upgrade.

I will move up to 1.068 to see if things change

Thanks

Paul

Hi all, @cheetah @paulg58 @vincent-mod ,

I think I know what the issue is here, I have experienced this myself.

Don't mind the detect above noise kappa value, it's not the cause of the issue I think.

Please try the following in 3) analyse stars:

Raise the star size area from 6 to 12 or even to 20

and then perform star analysis and registration.

Please let me know if registration is then fine on all frames.

If the star size area is too low, the brightest stars will not be detected/analyzed and this crucial for robust registration. So if you have stars that are bigger then 3-4 pixels Full Width at Half Maximum (FWHM), you will need to increase this star size area.

I will try to fix this for APP 1.069 in the coming days ;-), I will aim to make the star size area parameter redundant since it complicates things too much for the user.

Kind regards,

Mabula

I take it that the detect above noise kappa value and the "Raise the star size area" have changed in version 1.077 because I don't see it in the 3) analyze stars section.  I'm having issues finding stars in the images I have. This is most likely because of the near full moon. I can only see a couple stars in the image. It's pretty washed out. 

By the way. Great job on the program! I like how your program is set up with step by step work flow. Excellent! I did some image processing of the bubble nebulae and it came out nice. Very impressed.  I really enjoy using it. 

Thanks again.

Kip

Dear Kip, Mabula, Vincent, @mabula-admin @vincent-mod

I often encounter the same problem, with "star analysis failed" on most of my frames, although I can clearly see 5 to 10 bright stars in each frame, and certainly many more faint ones. If I can see them so obvious, why isn't the software able to extract them?
And why is there no possibility (any more) to fine-tune this process, e.g by "raising the star size area"?... typically the stars FWHM is ~8-10 pixels in my frames, and it seems that APP doesn't recognize them as stars.

As humans we are very good at identifying things like stars, even the faint ones. The issue for software is to be able to find them above noise levels and not wrongly identify them to avoid registration problems.

The "finetune" possibilities became absolete as the entire process was rewritten and made way more robust. I've not seen many issues anymore with the registration engine, it unfortunately usually comes down to the data that needs to be improved somehow for APP to do its work. I do wonder if other packages are able to process them, do you have any other just to check?