Alignment Issues

K

krphotogs

Well-known member
I am seeing issues where StarAlignment is not aligning the stars completely. I noticed this when after running ImageIntegration the resulting image is blurry, a star seems to have another duplicate beside it. I loaded all the registered files into Blink, and when I went through I noticed significant movement on a couple.

I recently added 50+ subs and all but 2 were good. I have tried altering the Noise Scale and Noise Reduction along with the Log(Sensitivity) which sometimes works, sometimes does not.

Anyone have any thoughts on this?

Kevin
 
Hi,

If you are seeing "double stars" then the StarAlignment had an issue. I had a similar problem and the issue was that my stars were not correctly focused. Have a look at my post "Mosaic of a Mosaic not aligning correctly"

Juan Conejero had this solution which worked perfectly:

The problem with these images is that they are out of focus. The new distortion correction algorithm implemented in StarAlignment requires accurate PSF fits for robust star detection. Unfocused stars have irregular distributions and flat profiles, which makes centroid determination uncertain. With its default parameters, SA cannot compute a valid local distortion model for these images.

However, you can tweak a couple of parameters to register these mosaics very accurately (to the accuracy that is achievable with out-of-focus images, that is):

- Set the peak response parameter to 1 (Star Detection section). This is necessary because the stars in these images are flat with relatively weak peaks, so SA's star detector must be more peak-sensitive than usual to detect some of them.

- Click the Edit Instance Source Code button (empty square, bottom right corner).

- On the source code editor window, look for the psfTolerance parameter. Its default value is 0.5 pixels:

Code:

P.psfTolerance = 0.50;

Change it to 2 pixels:

Code:

P.psfTolerance = 2;

- Click the Commit button (green check mark, top right corner).

- Set Registration model to Thin Plate Splines.

- Enable the Distortion correction and Local distortion options.

- Select the Ha image as registration reference.

- Apply the process to the OIII image.

- Use ChannelCombination with the Ha image and the registered OIII image.


spaceisnotblack
John
 
Ok - just tried that and it made it worse... dropped 2/61 frames that it would not align, but using Blink the ones that did there was a lot of shifting around. Thanks for the response though!
 
the other possibility is that there are a zillion hot pixels in your frames which is fooling staralignment. you can try increasing the noise scales and hot pixel detection controls in StarAlignment...

rob
 
Thanks Rob - tried that and seems better but still a lot of movement when going through with Blink. I will try increasing a little more. Appreciate your input.

EDIT: The integrated image looked good though.
 
Last edited:
I am late to the party... but if hot pixels are the problem... remove them first and then StarAlignment will work without modification (for reasonably focused stars). Have you tried this? It is all guesswork without seeing the images.
-adam
 
Thanks John - I went back and got a little more aggressive with cosmetic correction but there was not an over abundance of hot pixels. When I ran StarALignment again - 1 failed (ok). When I stacked the remaining 60 I saw a second star for most. Then I used Blink and found the 2 stars that were casuing my issue. Upon looking at them, one had pretty oblong stars (guess my tracking was off), and the second a little ovalness to the star. Maybe that was the issue?
 
Not sure if I am "John"...but certainly if the stars are doubled in the original data- then the issue is an image integration one.
Looking at the data or good images of the data to show what you are describing would be helpful.
-adam
 
So...I looked at the images. You are seeing the effects of different amounts of rejection. This is why some of the "doubled stars" do not appear doubled in your "better" integrated frame. You will note that on the right side the stars are "doubled" in both results... although a little less in the better one due to more rejection.

So the real thing to do is look *very* closely at your original data. It is still my belief this isn't about alignment. This deals with the fact that the images are doubled/smeared/multiple in single original frames. StarAlignment is trying to register on the centers of the doubled/smeared/multiple images. Then through rejection- some of the stuff does appear to go away. The trick is to use only the frames that have doubling/smearing in different places so that the rejection will be clean. This is allow you to use the maximum amount of your images.

I could be wrong...but without seeing the original frames- I can only give you my gut reaction based on my experience.

-adam
 
What you are saying makes sense. If I understand correctly, if any of the subs do not have rounded, focused stars in the same location then the StarAlignment is picking a center of the star that really is not, resulting in double images. So deleting those via blink would help. Would dithering have an impact on this? I think I need to do some research on how alignment actually works.

Sorry if these are basic questions, but spending a lot of time on Pix and trying to gain a better understanding.

Kevin
 
Sometimes during the capture of a light frame the scope might not guide 100%, which could cause stars to become slightly elongated. I guess that could cause the StarAlignment to be off by a pixel or three causing the double images. Doing the blink and reject is a good starting point. Then perhaps the image with the best FWHM and/or Eccentricity should be the reference image. Hope this helps.

John
 
I had the same problem and tried everything - after star alignment the calibrated, debayered, subframe-selector-weighted subframes were thrown in every direction - and they were well aligned to begin with (except for dithering).
What finally did resolve the problem for me was Noise Scales and Noise Reduction (under Star Detection) which I both set to 1 or 2 ? then all of the subs lined up perfectly!! ?
I suppose PI mistook all my hot pixels for stars - so now I suppose will have to retire my 13 years old astromodified full-of-hotpixels-Canon-5Dm2-camera and spend my mount-money on a Ra instead ?
 
According to my experience cameras don't age drastically. However, of course, a MasterDark cannot be successfully used forever.

Was your MasterDark prepared recently and correctly? Are you sure that the settings in ImageCalibration are correct? (See my guide , chapter 5, 6 and 8 for recommendations.)

If you can affirm both questions, try CosmeticCorrection after ImageCalibration in order to remove remaining hot pixels.

Bernd
 
You must log in or register to reply here.
Back
Top