Skewed Alignment

ngc1535

Well-known member
Feb 1, 2014
503
53
With regard to hot pixels- for some sensors these warm/hot pixels can accumulate charge in a non-linear way. Matching darks taking at the same usually takes care of things... but typically darks (or master darks) taken from the recent past are used and they will not necessarily perfectly subtract hot pixels. Changes in temperature, exposure time, scaling (optimization)... all can leave hot pixels behind.

You are definitely correct you need to remove them before debayering.
(which is what I demonstrated in the video)

You are also correct that hot pixels and other problems can be rejected during image integration. However... you can't get there if you cannot align the images first. If the stars in the image are bright enough- hot pixels will not matter much. But this is the reason I made the video. For many OSC and CMOS imaging situations where the exposure times are very short- hot pixels become important to reckon with.

Your hotpixels will not shift due to dithering. Only the stars (signal) will. The *will* shift after you have registered your images- which is why rejection will take care of them at that point.

-adam
 

fredvanner

Well-known member
Apr 17, 2019
86
4
Wells, Somerset, UK
I've traced my recent alignment problems to a file import setting left with the wrong value after some recent experiments. I have had these skewed alignment frames occasionall before; I'll try and find an example.
I have not used CosmeticCorrection. I always take darks, and I think they have always fixed any hot pixels. However, the video was interesting, and I now feel confident to use CC if I need it.
Failed dither; I had an embarrassing error with this. I now use PHD2 guiding integrated with APT to manage dithering, but before I learned how to do this I ran PHD and APT independently. I then made the mistake of selecting dithering in APT. APT dithered - and then PHD firmly restored the original guided position!
 

Farzad_k

Well-known member
Oct 15, 2016
178
1
With regard to hot pixels- for some sensors these warm/hot pixels can accumulate charge in a non-linear way. Matching darks taking at the same usually takes care of things... but typically darks (or master darks) taken from the recent past are used and they will not necessarily perfectly subtract hot pixels. Changes in temperature, exposure time, scaling (optimization)... all can leave hot pixels behind.

You are definitely correct you need to remove them before debayering.
(which is what I demonstrated in the video)

You are also correct that hot pixels and other problems can be rejected during image integration. However... you can't get there if you cannot align the images first. If the stars in the image are bright enough- hot pixels will not matter much. But this is the reason I made the video. For many OSC and CMOS imaging situations where the exposure times are very short- hot pixels become important to reckon with.

Your hotpixels will not shift due to dithering. Only the stars (signal) will. The *will* shift after you have registered your images- which is why rejection will take care of them at that point.

-adam

Adam, thanks. The darks I used are a couple of weeks old but match in characteristics to the lights. I will, however, keep an eye on them.

On the CosmeticCorrection and the adjuster, Hot Sigma and Cold Sigma: I have played around with them using your instructions. It looks like for the Hot ones, when I shift to the right, eventually a hot pixel in my view will disappear. If I push farther to the right more non-hot pixels will disappear. Can I trust that the Hot Sigma value that causes disappearance of the hot pixel in the view will remove all other hot pixels? Furthermore, how about cold pixels? All pixels appear the same in the calibrated subs which are monochrome (or appear to be ) at the time.

Thanks

Farzad
 

ngc1535

Well-known member
Feb 1, 2014
503
53
I think you mean to the "left." Smaller sigma values mean greater rejection of hot pixels. Yes, once you find a value that removes the hot pixel you are looking at...all others at the same sigma threshold (roughly the observed brightness above the background) will be corrected as well. So you want to use the largest sigma value you can that takes care of the majority of the hot pixels. The default of 3 sigma is really pretty good. 2.5 if pretty aggressive.

I have not used any sensors where "cold" pixels are an issue. I usually turn/leave this off.

-adam
 

Farzad_k

Well-known member
Oct 15, 2016
178
1
I think you mean to the "left." Smaller sigma values mean greater rejection of hot pixels. Yes, once you find a value that removes the hot pixel you are looking at...all others at the same sigma threshold (roughly the observed brightness above the background) will be corrected as well. So you want to use the largest sigma value you can that takes care of the majority of the hot pixels. The default of 3 sigma is really pretty good. 2.5 if pretty aggressive.

I have not used any sensors where "cold" pixels are an issue. I usually turn/leave this off.

-adam

Thanks a lot.