Horrible result with an integrated master light

[Diane Miller]

I'm using a simple Astrotrac and a Canon 7D Mk II with the Canon 400 DO II lens.  I shot the Orion nebula last April with a LP filter from a moderately LP area.  I got 35 sub and shot flats but no darks.  Each sub was 2 min at ISO 1600, and the stars were decently round but there was some drift during the set.  I followed the steps as best I could of calibrating the flats with a master bias (the master flat looked OK) and integrating, then calibrating the lights with the master flat.  Did CosmeticCorrection and Debayered then did SubFrame Selector.  There was some variation and I threw out the worst ones to leave about 25 good lights.  I registered them -- there was a little drift but I got no errors in the process.  Then when I integrated I got the mess shown in this link.  I've seen a similar thing with the strange trailing shapes in the Rejection High, but never this bad.  The drift went in about the 2:00 direction from the first frame to the last.

This exposure is blown out and I meant to do two others at shorter exposures for blending, but never got around to it.  So this one is just a learning experience.

I'm obviously doing something very wrong.  Is there any guessing what, without seeing the screenshots of all my steps?  I've run it 3 times with the same result.  I'm using the best steps and parameters I can glean from Jim Morse's cribsheet.

See the last 2 files here.  The last one is one of the raw files as seen in Lightroom.
https://dianemiller.smugmug.com/Photography/Files-for-PI-forum-2/n-BBs62B/i-3MKv3DD

Thanks for any help!!

[msmythers]

Diane

Have you used the Blink tool to look at your registered lights before final integration. You should see this problem. If it turns out to be a few subs then you can exclude them from integration. If i had to guess I would say you camera might have moved very slightly during the capture. Just a wild guess.


Mike

[pfile]

that looks to me like the images were not actually registered, or the unregistered images were integrated instead of the registered images. if you blink the ..._r.xisf files, do you see the stars marching along that line?

rob

[Diane Miller]

That looks like the problem.  Blink does show the "registered" files as not being aligned.  I ran it again to be sure I had done things right but maybe my parameters are wrong.  The screenshots are in this link:

https://dianemiller.smugmug.com/Photography/Files-for-PI-forum-2/n-BBs62B/i-cwnTnQX

Thanks for any suggestions what I have wrong.

[tdgm]

Have you looked at the reference image in the star alignment process as being the best for alignment??

[pfile]

hm, that actually seems fine. the only thing i can see is that there's a space in the folder name for the registered images. i think PI should handle that OK but as an old school unix person i never use spaces in filenames.

if you want to post the reference frame and one other frame i can try to register them.

rob

[Diane Miller]

Rob, I need to put you on retainer...

Took spaces out of folder names.  (Doesn’t make sense, but old hand here, too.  Why not.)
None of the raw files, viewed in Adobe Bridge, show any noticeable meteor or other trails.
Reran SubframeSelector again:
Noticed there was a little checkbox for Use cache.  Unchecked that.
Saw that I had an expression in Approval Expressions and a red X beside a Weighting, which I deleted.
Did a screenshot of the approval plot.  (See link below.)  Xed out 6 that were maybe outliers.
Used the third frame (right on the 0 line) for the ref frame in StarAlignment
Got convergence reached after 1 or 2 iterations on all the files – can that be right?
See the last 3 screenshots here.  Blink on those aligned files shows the same movement as before.

https://dianemiller.smugmug.com/Photography/Files-for-PI-forum-2/n-BBs62B/i-bZWGPgk

Something is wrong with my alignment.

[Diane Miller]

Have uploaded two of the approved files to my dropbox.  The lower file number is the reference image I used (third in the approved sequence), the other one is the last one in the sequence, with the most frame-to-frame movement, which doesn't strike me as huge.

Go to ftp://ftp.sonic.net/pub/users/elmiller

[msmythers]

Diane

It looks like you have a lot of hot pixels in these images. You might want to change the Hot Pixels parameters in Star Alignment. Star Alignment might be locking onto some of them in some of the images instead of just stars.


Mike

[Diane Miller]

Thanks Mike -- that could be it.  But I don't know what parameters I should change or to what...

This set was shot without darks.   Do darks help remove hot pixels?  (I have so much to learn here...)

[msmythers]

In the screen shot of Staralignment is the parameter. If you hover the mouse over the parameter you will see an explanation and suggested use. This is one of the great features in PixInsight I think. In so many ways it's better then formal documentation because it is usable without moving away from your work. Most tools have this feature.

As far as darks, yes that is one of the reasons to use them.

You will eventually get a good understanding of all of this. Your asking the right questions. There are no absolutes in this as a hobby for the amateur. Everybody has different issues based on equipment and environment. The settings that work for me won't necessarily work for someone else. Worse yet is settings that worked for me in July might not work for images captures in November. So many variables. But as you learn your equipment, your environment and your software it becomes easier and it becomes enjoyable. It's just takes time and practice with a good dose of patience thrown in.


Mike

[pfile]

yea - mike has it. with the default settings, there are > 2000 "stars" detected and most likely the majority are hot pixels, so SA dutifully finds the same hot pixels in the target frame and performs the null transformation.

i set noise scales and hot pixel removal to "2" and noise reduction to "1" and then SA found ~500 stars and registered the frames properly.

the calibration seems suspect - there's a strong "anti-vignetting" in these images. you can probably successfully remove it with DBE but you might want to check how your flats were calibrated. or maybe run without flats first.

rob

[Diane Miller]

OK -- Mike's on retainer too.  That change in parameters did it.  It found about 200 stars this time.  When I ran CosmeticCorrection maybe I didn't remove enough hot pixels.  I used Auto Detect and set both Hot and Cold sigma to 3.  Maybe I should try a Hot value of 2??  Guess I need to visually inspect some files after doing it.

Problem with the hints is they tell what the values do, but to the newbie they don't give much of a hint about "If this is wrong change the values to higher or lower", and give a hint by about how much.

Flats were calibrated with the master bias only, and I did not check Calibrate in that section.  (No idea what that means but that's what is on Morse's cribsheet.)

Many thanks for getting me this far -- I'll probably be back again...