Help needed registering images from 2 different scopes

[joelshort]

I'm trying to register with StarAlignmnet images from two different scopes/cameras and am having a little trouble getting all of them to register.  Out of about 200 images there are around 30 images that will not register with the settings I have tried.  I changed the registration model to 2-D surface splines, the RANSAC tolerance to 8.0, the hot pixel removal to 2 and noise reduction to 2 and that got me about half way there.  But there are still 16 images that refuse to register.  These images all look fine and are calibrated correctly.

Any further tips to get these images registered?

The next thing I might try is using CatalogStarGenerator to create a star field and match the images to that instead.

[frant]

Hi Joel,

I’ve found using previews helpful in tricky cases like this, defined over a smaller section of stars.
I’ve had cases where rotation was the cause of failure and manually rotating target subs to match prior to registration helped. Lastly, if I’ve managed to get one of the troublesome target subs to register I’ve trued using that as the reference for the remaining ones.

Hope that helps.
Richard

[joelshort]

Thanks for the tips Richard.

After messing around with the star detection parameters a little bit, the magic setting was Peak Response.  I lowered that from the default 0.8 to 0.4 and all remaining images registered.

[joelshort]

Well I spoke too soon.  I thought I had the images registered successfully but it turns out that the registration process distorted parts of some images.  Sometimes the middle 1/3 of the images morphs and the stars move.  Some images it's the outer parts of the image that morphs/distorts.  In either case the stars don't all match up and I get double stars in some areas of the stacked image. 

This is proving to be quite impossible!

I am combining images from a 10" RC at f/6 (pixel scale 0.81 arcs/px) and a 80mm refractor (pixel scale 1.62 arcs/px).  I have tried registering to 1 of each sets of images.  I have tried using CatalogStarGenerator to create a star field and register to that.  The false star catalog worked best but even that wasn't perfect. 

I'm at my whit's end and I just can't seem to get these images registered properly.  Any help is much appreciated.

[GJL]

Stack the two masters. Then register them with Star Alignment, when distorted better Dynamic Alignment, perhaps twice. Finaly addition with PixelMath.
Hope this helps.

[pfile]

another diagnostic is to use StarAlignment in "detected stars" mode to see just what SA thinks are stars in the problematic images. if it's detecting 1000s of stars it's probably still picking up noise and hot pixels. although generally when that happens the alignment transformation is null because the hot pixels are in the same place on every image... still it is worth seeing what is going on.

GJL's suggestion is good.

another idea might be to drizzle the refractor images and then they will effectively have the same pixel scale as the RC images.

rob

[joelshort]

Thanks for the replies.  After a first pass, it looks like stacking the individual channels and then registering the masters together is working. 

I'm sure that previously I was picking up noise, so I increased the noise reduction in SA and that helped when I was trying to register ALL the subs to the same reference image.  But now stacking each channel and registering the masters is working much better without any noise reduction in SA.