Image registration

[gregbradley]

Recently I suddenly have had trouble with CCDstack and the CCDIS registration plug in. After faithful and accurate service it now is not registering all images properly. I have recently installed CCD Inspector and I have noticed before if you have the trial version of that installed the CCDIS registration plug in can act as if its in trial mode as well.

Is anyone using PI to register their images?

I tried but wasn't sure of how to use it. I take it you use the Star Alignment Process.

Greg.

[RobF]

Very powerful routines Greg, definitely worth getting to know.

Possibly too much info here:
http://pixinsight.com/doc/tools/StarAlignment/StarAlignment.html


More pragmatic here: http://www.deepskycolors.com/PixInsight/ImageRegistration.html

[RickS]

Greg,

I use PI for everything starting with calibration and registration. To do this combination the easiest way is to use the BatchPreprocessing script. If you just want to do registration then StarAlignment is the tool of choice.

The FITS format has some weaknesses as a standard so you'll find there are issues with moving files between applications like CCDStack and PI. These can be resolved (PI has a lot of flexibility if you mess with the preferences) but it is generally more difficult to skip between applications.

Cheers,
Rick.

[Moon]

Greg
I've been having troubles with CCDIS recently too. I wonder if something broke in a recent upgrade?
James

[gregbradley]

Thanks very much for those links. I'll work my way through the 2nd one.



I was using that but I must need to click on something as I could not get it to work. PI is no doubt a powerful program but its not user friendly which is a shame as that would really top it off.



I wonder if I have an earlier download I can use. I also notice the latest update of CCDstack 2 is bugged. It now often says it can't read files that it routinely used to.

Updating software is not always a step forward.

Greg.

[gregbradley]

What the specific problem I am having is I use the Star Alignment tool. I click on the various images and the reference image. I insert an output location and the file type.

But when I execute the process and I look in the output directory there are no files and it simply makes a copy of the luminance image with the word registered added to the file name but the luminance image is the reference image so why would it be changed?

So how do you set it to save the registered images (LRGB registered to L).


Greg.

[RickS]

How did you "execute" the process, Greg? Sounds like you might have applied it to a window with the Luminance image (drag and drop the triangle process icon). If you want it to process the list of target images you need to run it with "Apply Global." You do that by clicking on the little circle icon at the bottom left of the process.

Cheers,
Rick.

[gregbradley]

Ahh. I think this may be the problem. Tks Rick. I'll try that now.

Another minor issue. The update repositories menu item is not working. It shows the updates ready to be installed but when I click ok it comes up with an error "network error".

Greg.

[RickS]

I haven't seen that one before. You're obviously able to talk to the update server or you wouldn't get getting the list of updates. You mentioned error code 404 on your post on the PI Forum. 404 is a HTTP "not found" error, so I guess a file/page on the update server is missing (temporarily or permanently.) Hopefully, one of the PI Dev guys on the forum will be able to help. If I could reproduce it here I'd have a go at debugging the problem but I can't...

Cheers,
Rick.

[RobF]

Remember too Star Alignment is only going to register your images ready for stacking. You need to then use Image Integration to stack them, via whatever algorithm and settings you choose - probably the most important part.

Make sure to check out Rick's earlier suggestion re Batch Pre-processing script (from script menu). Is much more efficient way of operating, although for optimal performance running Image Integration on the calibrated and registered images is still recommended. (pah! :lol:)
(Might overcome a lot of your user friendless concerns......)

[gregbradley]

Thanks Rick. I'll post it on the PI forum. I worked at one stage and then stopped for some unknown reason.




Thanks for that. I did end up getting the registration to work and it did a good job, better than what I could achieve in CCDstack. There were some dodgy tracking subs in the green and so the star shapes are what were throwing the alignment routine off.

I'll have to try PI for integration. I think I need to watch the Warren Keller tutorials I bought some time ago to find my way around the basics. I did not even know about the global button!

PI has lots of interesting tools and I would like to expand my processing routines.

Greg.

[RickS]

ImageIntegration is very powerful, Greg. If you do an initial integration with no rejection you can determine a best case noise reduction target (the median noise reduction number printed on the Process Console.) Then you can play with rejection algorithms and parameters to see how close you can get while still rejecting hot pixels, cosmic rays, satellite trails, etc. It's a bit time consuming but certainly squeezes the best out of your hard-earned data.

Cheers,
Rick.

[gregbradley]

Thanks Rick.

I'll be sure to give it whirl.

Greg.

[sil]

+1 if you've got lots of noise and/or distorted/streaking stars these tools can fail. In difficult cases i use the Dynamic Alignment tool to register the first five frames then integrate them to give me a cleaner image I then use as my source target image for StarAlignment or batchpreprocessing to register all the frames against. Or in worst cases i just use DA for the whole registration process. Before all this I tend to run the Subframe Selector script to find the worst frames (where the stars are most distorted) and weed them out first so I'm not fighting vibration wiggles in the source.

[Paul Haese]

In CCDstack2 Greg you might try using star match. I found it very accurate and you can ramp up the amount of stars to use. CCDIS seems to work best when there are not heaps and heaps of stars in the field.

[gregbradley]

Thanks for that. In the end PI using Star Alignment worked best so I got through it! I think a few subs with elongated stars was throwing it off.



Thanks Paul. Star Snap? Yes I ended up using that and it worked better than the CCDIS plugin but still not 100%. PI ended up working the best. Still not 100% ideal but close. Life is always harder in AP when the basics are not done super well like perfect autoguiding etc etc.

Greg.