HOWTO: Data from multiple sources?

[mads0100]

Hey guys,

   Does anyone have a good tutorial on how to combine data from multiple sources?  We're looking at doing a collaboration over at CN and we'd like to take data from a variety of image scales.  I understand that Registar is the best option... but the majority of us are Pixinsight users. 

Any tips, techniques, ideas would be most welcome.

Thank you for your time,

Chris

[pfile]

hey chris, it's really simple… just register all your images to the same reference frame and PI will rotate, translate and scale the images to match.

no need for other software...

rob

[Juan Conejero]

... data from a variety of image scales.  I understand that Registar is the best option...

Are UFOs real? What makes you think so? 

The practical aspects of this question are really simple as Rob has said. From a more technical perspective see this tutorial:
http://pixinsight.com/tutorials/sa-distortion/

Anyway, should you have any problem with image registration in PixInsight, let us know and we'll solve them.

[mads0100]

Thanks Juan/Rob, I didn't realize it was so easy!

Regards,
Chris

[Tromat]

Hi,

In my opinion the registration part is trivial, the real challenge is the integration of many different subs, with different SNR and different FWHM. I personally didn't get any good results with this kind of exercise, the best images are always weakened by the worst ones. I guess the key to obtain a good result is the normalisation but I don't think there's a way to get a perfect result. I hope I'm wrong though.

[pfile]

yes i agree - if the FWHM of the data sets differs too much it will be very difficult to get good results. but if the systems involved are similar it should work out.

rob

[mads0100]

We ended up doing it.  The thread is located over in the Beginner forums.  Our image scale varied between .7 and 1.8 (basically some guys with RCs and some guys with ED80 type scopes).  I was definitely surprised at how well it turned out.  The big difference between my image and Madratter's was that I created a synthetic luminance and combined it with the real luminance's we had.  Unfortunately, we had a ton more RGB than luminance data so it was necessary to get a good result. 

Here is my example using 32-ish hours of data:

[pfile]

cool. looks pretty good.

along these lines, there was a reddit m42 project going on similar to this but i'm not sure what happened. maybe josh got busy.

rob

[mads0100]

Maybe.  I never saw any results from that.

Honestly, it was kind of disappointing to see the results.  Most people ended up skipping data sets or decided later on they just wanted their own data over the groups for some reason.

I thought it was amazing.  I wish more groups did this.

Take care,
Chris

[chris.bailey]

Chris

We managed to amass some decent data over on the UKAI forum. This was from 4 very different setups (reflectors and refractors) with image scales in the range of 1.5-3.5 and I was pretty pleased with the way Star Alignment with Distortion Control pulled the data together. I did use the sub-frame selector script to pick the best 75% from each data set.

Chris

[mads0100]

Chris, that's beautiful.  Thanks for sharing!

[chris.bailey]

Chris

Thanks, I expected more people to jump at the chance of pooling data but it took the best part of a year to amass sufficient data. Add in the problems of different peoples approach to framing, calibration and even getting good focus and its one of those things that seems like a good idea until you do it. Getting 24hours of data on a single subject under UK skies and with UK weather though, does make it appealing.

Chris