Large Scale Mosaics

[Haverkamp]

Okay Tom, good to hear ;-)

“What do you mean by the RMS screen shot of the registration?”

After registration of the mosaic is finished, in the bottom file list window, there is a column with registration RMS and the #stars that were used in the registration. That information could help me in improving this. (Possibly, due to the amount of panels, the registration path to the reference becomes too long, but this is rather technical.. )

[O’Donoghue]

Those settings helped on the re run. I m posting the 5×5 image. There is only one area on the right hand side of some star misshape, and one area where there is flaring to the top right of the Pacman nebula. I’ll be able to fix these once stretched.

This is just a simple histo stretch and a small contrast tweak for the screen.

Tom

[Haverkamp]

Excellent @todono1x! Do you know which settings exactly gave you the best result?

I am focusing now on speeding APP up, that has higher priority now….

But I am confident that I can apply some more improvements in the registration engine for mosaics like this. So it’s on my list to try to improve this further.

[O’Donoghue]

I ve been trying the triangles and it just hangs up. Not sure why but lots of Array errors.

The settings you suggested in your last mail worked best, but at a lower star amount 10000 was too much and it hung. I think I used 7000 in the end. I have  no RMS values for you at this time. I would need to re run the experiment.

[Haverkamp]

Hi @todono1x,

Okay, I should do some more testing with these extreme settings to see if I can trigger the Array errors you mention.

I’ll see if I can make the registration engine a bit smarter. Meaning, if quadrilaterals don”t work, automatically try triangles..

For such a big mosaic, I think I can improve the distortion models a bit more to get even better results. I will work on that after the priorities that I now have which are on the RFC.

[Author]

Posts