Mosaics and Keeping It Rectilinear

[avandonk]

Because the sky is a sphere whenever we produce an image we are mapping a spherical surface onto a flat one. Good optics such as an FSQ106 or Canon super telephoto do this better than a fraction of a percent.

We have a problem when making extensive mosaics as the small distortions of the spherical mapping onto the flat plane of the reference or starting image will badly distort any images that are registered by software such as RegiStar. This effect gets worse the further the frame is from the reference image.

The first image below is a mosaic made with Registar using the image with the black edges as the reference image. It is obvious that the frames distort and the corners are no longer square.

One solution is to use an image of a very good rectilinear lens with a shorter focal length and use this as a template.

There is an easier way. I use Starry Night Pro to produce an image to use as a template. SNP does a very good job of generating a rectilinear image or projection. Even when made into mosaics to get a bigger field.

The main trick is to upsize the SNP image to match the resolution of your optic and camera and use RegiStar with the noise compensation to eliminate any possible artefacts.

So again the first picture is is a mosaic made with Registar using the image with the black edges as the reference image.

The second is a mosaic made with SNP image as a template. Note how all the frames are nice and square.

The third image is the SNP image.

Hope this can help someone with their mosaics. It has certainly made this tedious process easier for me.

Also SNP is cheaper than a high quality rectilinear lens!

This mosaic has thirteen panels to eliminate reflections from the bright stars of Crux. And people wonder why I sometimes get frustrated and a bit grumpy.

Bert

[h0ughy]

Fantastic information Bert. Will note for future reference

[avandonk]

Glad to help a bit David. This imaging is difficult enough and often I find myself groping in the dark for solutions to problems. So when I can contribute something that I find works, it is published here.

Bert

[gregbradley]

That's very good of you to show that valuable information.

A great piece of knowledge.

Greg.

[multiweb]

Cool tip Bert. :thumbsup:

[DavidU]

Thanks for showing us. Your problem solving skills show through here.

[allan gould]

Bert
Thats a great solution for mosaics. I'll give it a go at Astrofest this year.

[RobF]

Yes, thanks again Bert for helping along the rest of us - even if we are a couple of k's behind down the road :)

Just out of interest, Pixinsight now has a tool for generating artificial starfields for just this purpose from Star Catalogues - works very well too for dealing with these geometrical issues.

[avandonk]

Here is a large image of the template I made using this method.

16MB

http://d1355990.i49.quadrahosting.com.au/2011_03/templatefinal_.jpg


Note how all the panels are the same size and square (rectilinear).

I used the 120s stacks without any attempt at calibrating etc as I wanted to see the edges of the panels.

With twelve panels I have 12 x (6 x 20) = 1440 images to process. That is about 280 GB of tiff data. Something to do on cloudy nights.

Bert