Funny numbers for image 41

[KeesScherer]

“Panel 41” is a single exposure light frame that looks good visually with a slightly higher FWHM but why does APP compute these rediculous numbers for star density, quality score and registration RMS? It is the top row in this screen print.

[Haverkamp]

Hi @keesscherer,

This is a clear indication that there is a registration problem for this frame for some reason…

Can you register this frame without problems to one of the frames that has overlap with this frame?

Does this error repeat itself with different registration and/or star analysis settings?

Mabula

[KeesScherer]

There is a problem with dynamic distortion correction. When i only use panels 41+42 i get the Java error as shown in the screen print. When i swith the registration model from projective to calibrated projective the java error does not show. The resulting registration RMS is now 2.35, so still too high and there is some mismatch as you can see in the second screenprint.

[Haverkamp]

Hi @keesscherer,

Can you send me these two frames so I can investigate what’s happening here? Looks like something in frame 41 is different than in the other frames… but I would love to check it thoroughly.

Did you use default settings in star analysis?

Mabula

[KeesScherer]

Uploaded 41 + 42 to dropbox. The strange thing is that i made a mosaic of frames 41,42,47,48 without a problem 2 days ago. I used star an. settings: 4, 0.1 filter on kappa 4 limit to 1000 stars.

[Haverkamp]

Hi @keesscherer,

Thank you, I have found the bug. It happened in

-mosaic mode

– not same camera and optics

– dynamic distortion correction

– projective model

The bug caused to start the regression calculation with wrong initial parameters, in this case the “wrong” parameters couldn’t be found. So a very good bug to solve ?

Furthermore, I see another and even more serious bug in the star analyser when “filter star profile” is enabled with your data (first time I encounter this.. but must have affected other data processing as well):

minimum star size 4

clip profile 0.1

filter on

kappa 8

limit stars to : doesn’t matter, screenshots with 1000 stars

It’s a serious bug, so very happy to have found it. 2 screenshots:

first with “filter star profile” off, the other with “filter star profile” on. The screenshots show the “details” of “star map image viewer mode” on panel 41.

Notice that with filter on, there are stars found with FWHM max, min of 1,00 & 1,00. These are bogus results, the star locations are rubbish as well… This must seriously affect registration in a bad way.

[Haverkamp]

@keesscherer,

solved the error in the mosaic algortihm and also the error in the star analyser. Should improve results and robustness in registration. Will compile a version with these fixes later today, so you can test with your cygnus mosaic.

[Haverkamp]

Hi @keesscherer,

If you login at astropixelprocessor.com, you can download a new linux DEB version, 1.040.

I have fixed the 2 problems, and I have worked on the DEB installer. Please let me know if this works okay on your Linux system ;-)

Mabula

[KeesScherer]

and I have worked on the DEB installer. Please let me know if this works okay on your Linux system

Nice, APP now shows up in the Science & Maths menu!

[Haverkamp]

Yes, still need to fix the application icon in the menu though ;-)…

[Author]

Posts