Star alignment failure, galaxy image with few stars

[astrorunner]

I'm really struggling and can't get StarAlignment to register any of my subs of the Sunflower galaxy. I'm using a small sensor camera (ASI 290) and a long focal length telescope (C9.25) resulting in an image with virtually no stars, and that seems to be causing the problem (never had this problem with any of my other processing, but they all were subs with stars).

I've tried suggestions from a processing tutorial book to increase RANSAC up to as high as 8 and I've lowered Log(sensitivity) all the way to 1. I've also changed noise reduction and noise scales to see if those made any difference (they did not). If downloading anything from the Console will help let me know and I'll display here.

Doing some internet searches I saw where people were having success with lunar registration using "FFTRegistration". This did nothing for me either, seemed like it computed shifts in x and y that were not large enough, even with the option checked for "enable large translations".

Here's a link to the first 20 calibrated subs from my data (I have several hundred in total): https://drive.google.com/drive/folders/0B2wyjA4jhK9VQkZ6aDhxbVRwVHM?usp=sharing

You guys have been great before and I really appreciate your help, I'm anxious to stack these. Thanks!

[John_Gill]

Hi,

Try the following:  In StarAlignment ---> Star Detection
   Detection Scales increase to 6 or 7
   Noise Reduction increase to 2 or 3 (I have used up to 5)
   Log Sensitivity reduce to -2 or less
   Peak Response increase to 1.0

Look up
John

[astrorunner]

Thanks John. Those settings did help, it seems very particular. I'm getting about 70% registered now. I've also found that doing cosmetic correction after calibration (with batch preprocessing) helps. There are a few other settings I'm still experimenting with too but I'm not far enough along to post results.

[astrorunner]

I wanted to just finish up this thread by saying I found some settings that worked pretty well, thanks to the help of another friend who read my post.

I ended up with:
Detection Scales 8
Noise Scales 2
Noise reduction 8
Log (Sensitivity) -3
Peak Response 1.0
Maximum distortion: 0.27
RANSAC tolerance: 8
RANSAC iterations: 2000
Descriptors per star: 20

These settings resulted in about 95% registration.