[Solved] Random banding in registration reslts

Hi Darrel @lazyimager,

Thank you very much for your question/issue. Please accept my apologies for the late reply to your message.

That registered view of your data when drizzle is enabled, actually is perfectly fine. This is exactly what drizzle does when the data will be integrated, data is drizzled down on the new target pixel grid and not all pixels will be hit depending on the droplet size, drizzle kernel and scale factor. In development, I decided to show this clearly to the user when the registered view is done.

If you continue and integrate the data with (Bayer/X-trans) Drizzle, you will actually see that all is okay. Did you try?

Mabula

Hi Mabula,

Yes I continued and as you predicted all went well. It's been awhile and I don't recall if I was having some other problem that caused me to look at the registration step, but nonetheless, not having any problems now. I have tested several images comparing normal vs drizzle integration and so far don't see much difference. 

May I ask you a related question? When I examine my subs after I Analyze Stars, I typically see FWHM Min-Max in the 3.5-4.0 range. Are these statistics reliable indicators for determining if I am over/under-sampling? For example, if I am imaging at 1.99"/pixel and the FWHM statistic is 3.5-4, I would think I am perhaps slightly under-sampling, assuming the "proper" sampling rate is no more than 1/2 FWHM, i.e., 1.75-2 in this example. Not trying to get too hung up in the theory, but am curious if I am interpreting the Analyze Stars output correctly and my understanding of sampling is sound.

Thank you and AstroPixelProcessor rocks!

Darrell

Hi Darrel @lazyimager,

Great, thanks for the feedback.

May I ask you a related question? When I examine my subs after I Analyze Stars, I typically see FWHM Min-Max in the 3.5-4.0 range. Are these statistics reliable indicators for determining if I am over/under-sampling? For example, if I am imaging at 1.99"/pixel and the FWHM statistic is 3.5-4, I would think I am perhaps slightly under-sampling, assuming the "proper" sampling rate is no more than 1/2 FWHM, i.e., 1.75-2 in this example. Not trying to get too hung up in the theory, but am curious if I am interpreting the Analyze Stars output correctly and my understanding of sampling is sound.

THe FWHM Min-Max indicate the FWHM value in pixels. They should be reliable if you accurately know the imaging scale as you indicate.

So with an imaging scale of 1,99"/pixel, it means the FHWM of your stars is on average 7-8 " (arc seconds), right?  That actually means that either your optics or your sky conditions produce quite big stars with 7-8 ".

Regarding under- or oversampling. With sampling of 3,4-4 pixels, you are actually oversampling and not undersampling here. If the FWHM would be around 2 pixels, you will start to undersample and drizzle really starts to make sense.

So it is good that you ask this 😉 currently, you are already oversampling I think and using drizzle should not improve sharpness really, or does it?

Mabula

Thank you for your helping me better understand this. Depressing to understand my stars are even worse than I thought. And no, drizzle integration does not improve my images.

I am running a Celestron C8 with Hyperstar V4, which images at 1.99"/px at F/1.9. Collimation, backspacing, focusing and cable management of cables in front of the objective (which causes odd diffraction spikes) are all very tricky with this setup. I have gotten better, when I first started I was getting FWHM values in the 4-6 range. At least I've gotten them down to 3.5-4 now. 

Thanks for you help,

Darrell

One more question if I may. The FWHM Min-Max. Is this describing the stars with the lowest FWHM versus stars with the highest FWHM, or is it describing the average minimum width versus maximum width (i.e., the eccentricity) of all the stars sampled? Hope this makes sense.

Darrell

Posted by: @lazyimager

↑

Thank you for your helping me better understand this. Depressing to understand my stars are even worse than I thought. And no, drizzle integration does not improve my images.

I am running a Celestron C8 with Hyperstar V4, which images at 1.99"/px at F/1.9. Collimation, backspacing, focusing and cable management of cables in front of the objective (which causes odd diffraction spikes) are all very tricky with this setup. I have gotten better, when I first started I was getting FWHM values in the 4-6 range. At least I've gotten them down to 3.5-4 now. 

Thanks for you help,

Darrell

Okay, I fully understand. Please realize that the sky conditions play a major role in what resolution you can obtain with deep sky imaging / long exposure astrophotography. At most locations, the seeing of the sky will be between 1-5 arc seconds. 1 "being extremely good (dessert or high in the mountains for example) and 5" being very bad conditions. So with FHWM of 4 pixels being 8 arc seconds, I think the main issue will be the coliimation of the telescope combined with all adapters in your optical train for now.

One more question if I may. The FWHM Min-Max. Is this describing the stars with the lowest FWHM versus stars with the highest FWHM, or is it describing the average minimum width versus maximum width (i.e., the eccentricity) of all the stars sampled?

The FWHM max and min size represents the median star that was analyzed. So if you have perfect round stars on average, the min and max will be the same. So min and max tell you how the shape of your stars is. So for example, if min FWHM is 3 and max FWHM is 6. It means your stars are elongated where the fwhm profile is 2x is large in one direction compared to the size in the perpendicular direction. So it is like the eccentricity, where I think that the eccentricity is a very mathematical way (not very intuitive i think ) of describing it.

Load an image with stars into the image viewer, and set the image viewer to the star map mode, then the profile on the right side will show this. This reported FHWM profile with min and max size represents the median star in all the stars that were analysed.

Mabula

Thank you for all these explanations. I appreciate it.

Darrell

You are most welcome @lazyimager 😊