Integration hangs.

[erikgu]

Hi !
I updated to 1.7.5 yesterday. The program hangs when i am doing an ImageRegistration, StarAlignment.  10 out of 14 images seems to be registred, the program hangs when processing the 11. I have done this several times wit, images are processed when limiting to the 2-3 "last" images using the same as reference images, but the integration of all images gives a bad result. The latest test has been executing for about 10 hours.

My computer is a rather powerful windows pc with 12GB Ram and Windows 7.

Attatching processing consol log.

Erik G

[Juan Conejero]

Hi Erik,

This log shows that StarAlignment is havind difficulties to register these images. It manages to register 10 of them (in several tries), but it fails to register image #11.

Without having a look at at least three of these images I can't know what hapens. Could you upload images # 1,  3 and 11 somewhere? In this way I could analyze the problem and hopefully find a solution.

A further question. When you say 'it hangs', what do you mean exactly? The term 'hangs' referred to a computer usually means that an application enters a lost state where it doesn't respond and you have to terminate it in a forced way. Is this what happens? Or is it just that the StarAlignment process fails?

[erikgu]

Hi !

With hangs i mean that the Registration process hangs, the computer itself is ok. I can be terminatet by the Break button in the Processing Console.

Link to images, hope you can get them from there, sees that the download is slow.

http://dl.dropbox.com/u/59585830/Pixinsight%20Intg/debayer_IMG_1124_c.fit

http://dl.dropbox.com/u/59585830/Pixinsight%20Intg/debayer_IMG_1126_c.fit

http://dl.dropbox.com/u/59585830/Pixinsight%20Intg/debayer_IMG_1137_c.fit


Erik G

[Juan Conejero]

Hi Erik,

Thanks for the images. These images have too much noise for StarAlignment's star detection routine. SA detects thousands of noise structures as false stars. With default parameters, SA detects more than 30,000 'stars' on each image, which obviously are not real ones, and this leads to an overwhelmingly large amount of false star matches. Even under these conditions, SA is able to align most of your images, which is a miracle.

The following parameters (Star Detection section) perform an extremely aggressive noise reduction to the image seen by the star detection routine, and decrease the sensitivity of the detector to make it more robust against noise:

- Detection scales = 5
- Noise scales = 4
- Hot pixel removal = 2
- Log(sensitivity) = 0
- Peak response = 0.3

These parameters improve registration of your second image (using the first one as reference), but even with them, your third image cannot be registered. I'm sorry.

You definitely need to improve the signal to noise ratio. What exposure times did you use?

[erikgu]

Hi.

This is the first time i an doing the calibration and integration in Pixinsight. Have to check the steps to see if i have done something wrong.

Erik g.

[Nocturnal]

Hi,

it's not that you did something 'wrong'. Your images are very noisy, probably because they were taken at very short exposure lengths. If the exposure times were 30 seconds or longer then perhaps your camera isn't configured correctly.

Please upload the raw files next time rather than these huge 32b float monsters

[erikgu]

Hi !

The images have 600s exposure time. I have looked through the camera and filters and it seems to be a very dusty filter that can have caused the bade images. Here is a link to the raw images. Would have been nice if someone could take a look to see if my thoughts could be right or if my flats/darks could have caused the bad calibrated images

http://dl.dropbox.com/u/59585830/Pix2/IMG_1124.CR2
http://dl.dropbox.com/u/59585830/Pix2/IMG_1126.CR2
http://dl.dropbox.com/u/59585830/Pix2/IMG_1137.CR2

Erik G

[oldwexi]

Hi Erik!
Tried to check your "1126" image.

The raw CR2 file looks perfect!
Your calibratet 1126.c file is unusable in my opinion.
Either  the dark or bias or flat do not fit to the raws. The calibration destroys your raw file.
Check the histogramm of your raws and calibration masters.
i assume something went wrong with your master files.

Your fine Raw File with ABE flattened is visible here:
http://dl.dropbox.com/u/57910417/img_1126_ABE.jpg

Aloha
Gerald

[erikgu]

Hi Gerald !

Thanks for inspecting my image. Good to hear that the raw seems to be good. I am, as written in a previouse item in this post, stacking an calibration in Pixinsight for the first time. It is also the first time i have taken flats. I have followed the proposed workflow but might have done somthing wrong in the flats/darks calibration steps.

Might be a stupid question but did you debayer the raw before you did the ABE ?. I tried ABE with default setting but that failed, i think the error message indicated that it could not model any background. ABE was sucessfull when i debayered the image but the got a green/brown color. Not like the one you had a link to. 

Any hints ?

Erik G
 

[Juan Conejero]

Hi Erik,

Any background modelling process, such as ABE and DBE, needs debayered data.

When you apply ABE, you have to select Correction=Subtraction in the Target Image Correction section. This should neutralize the background of the image as a side effect.
   

[erikgu]

Hi i have tested Debayer and ABE but the images becoms green. The DSL_RAW setting is as proposed in the DSLR workflow. ANy idea what i am doing wrong ?

Erik G

[erikgu]

Hi forget last reply

Erik G

[Alejandro Tombolini]

Hi Erik,

Most of the time they look green after debayer.
Try apply and Auto STF without linking RGB channels and see if it look fine.
If so you will need to apply BackgroundNeutralization and ColorCalibrations routines

Saludos, Alejandro. 

[erikgu]

Hi !

Thanks to all. Time to go back to see what went wrong in the Calibration/Master creation steps.

Erik G