Newby can't register frames

[kor]

Hi all,

I can't seem to get the Star Alignment tool to actually register the frames that I give it.  I am new to PixInsight, so I'm probably doing something stupid or missing something trivial, but if anyone can give me a pointer, I'd appreciate it.

After calibrating and debayering these frames:  (Warning: Large  files to download)
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0008_c.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0019_c.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0032_c.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0040_c.fit
I run the Star Alignment Tool, defaulting everything except the Reference Image, Target Images, and Output Images.

The "8" frame is the reference.  For the other three frames, >2000 stars are identified and around 1600 are matched.  The Transformation Matrices get some values changed but the "rotation", "dx", and "dy" all end up being zero and resulting frames are not registered:
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0008_c_r.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0019_c_r.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0032_c_r.fit
http://www.kor-astro.net/PI_Support/001/debayer_NGC%204565-1_light_0040_c_r.fit
Though the MD5sum of the registered files is different, the image data appears to be unchanged--If I drag a registered image on top of the corresponding unregistered image, it lines up perfectly.

Here is the console log from the Star Alignment tool: http://www.kor-astro.net/PI_Support/001/Star%20Alignment%20Log.txt

The frames are part of a larger set and were taken over a period of 3 nights.  Each is dithered and slightly rotated:
Frame 8:   359.31 degrees
Frame 19: 359.84 deg
Frame 32:     0.36 deg
Frame 40:     0.37 deg

I have read the Star Alignment tool's documentation page, used Working Mode: Matched Stars to verify that it is finding good stars, view several tutorial videos and read thru many of the processing examples.  I'm probably just missing something obvious.  I ask for your help.

I am running PixInsight v01.08.01.1092, with updated through Jun 4, 2014 applied.  OS is MS Windows 7 Home Premium SP1.

I can't think of anything else to add.  If you have any questions or need more data, please let me know.

Thanks - Shane

[kor]

OK, I think I've got it:  Way to many hot pixels being detected as stars.

I ran the frames through CosmeticCorrection:
   Enable Use Auto detect
      Enable Hot Sigma @ 3.0
      Enable Cold Sigma @ 3.0

Then I did StarAlignment again changing the following from the defaults:
   Star Detection
      Noise scales: 2
      Hot Pixel removal: 2

Now it finds about 250 stars and matches about 20 in about 300 RANSAC iterations.
The rotation, dx, and dy are non-zero and the registered images line up!

Can some expert tell me whether this is a reasonable solution or not?

Thanks - Shane

[pfile]

yes i think that's a good solution. if you just get rid of the hot pixels that might be enough. i've never had star alignment fail on me after good calibration/cosmetic correction.

rob

[kor]

Thanks Rob,

Turns out most of my dark frames failed to scale properly.  I'm going to get some new dark frames this weekend and try again.  My original's were long as I was some other package.  Now that I am using PixInsight, I need to get them closer to the image time and temp so, hopefully, the optimize algorithm will use them.  I think when I get that working, my hot pixel problem will be greatly diminished.

I just got done with an initial processing.  Even with 42 10 minute frames, it's noisy and lacking in color.  I think I need to improve the calibration and try again.

- Shane

[bitli]

If you took darks with another sofware than the current one, check that both have the same orientations.
If you blink a dark and an image you should see that most hot pixels align.  If they do not, mirror the image (upside down) and see if if matches now.

-- bitli

[kor]

Thanks Bitli,

The darks I was using were taken with the same acquisition program, but were intended for use with a different processing program, which recommended that scaled master darks be at least 5 times longer than the lights.  So my master dark was built from 1 hour subs, since I usually take 10 minute light subs.

I am in the process of redoing my darks with 10 minute subs and, if I keep the temperature stable for another two hours, I'll try again with those.  Even if they don't work for PixInsight's optimization algorithm, I can still use them directly, non-optimized.
 
- Shane (rapidly learning more and more about PixInsight!)