I use registar a lot and I think it's great for registration. Today I tried something new. I got two highres files one in Ha and one RGB and tried to register them. The start field is extremely busy, the shots didn't even align and they were taken with two different scopes, so different image scales. In these two particular files even CCD Inspector couldn't read the starfield for collimation or field curvature data, I assume because it is so dense.
Here's the results. It did a good job in the middle but I was amazed to see the "drop" patterns it generated at the top and botton of the pictures. So it does a lot more than rotating and scaling pictures, Looking at the pattern it seems to me that it's looking at neighbouring stars for each registered star and warps the surrounding accordingly then move on to the next one in an iterative way. That would explain those artifacts and patterns.
What do you think? Groovy alignment hey?
Just checking for second opinions and making sure you see the same thing so I can rule out what I'm drinking right now...
