Thanks Schmeerlap,
your kind words show me that you would like to see my clipfile.
Blackhole, look again, there is only one version of this image yet and the colours are in order. Although I have learned since that secondary spectra differ in some other ways, but anyone can adjust his rainbow as he likes, so I won´t change much now.
dandelO, Thanks for the congrats.
I am not surprised that you get the clue, maybe I would have tried it with a bunch of spherical distance shaders if I had not seen your metaclouds, that would have been a pain.
So, I used one planets atmosphere with 14 cloud layers.
Funny thing is, I thought I would have to use the intersection of this sphere with two linear distance shaders to cut a slice out of it, but I realized that those shaders weren´t plugged in only after it looked like it does.
There is only a diffuse shimmering of colours that shows it is a sphere, not an arc, and I think the sphere-shape helps blurring the banding, at least where the density is high enough.
I am working on a second version now that I will post tommorow together with a clip file.
But I still dont understand why the violet arc seems to be flatter than the others.
Maybe it is foreshortening and will be solved if the whole thing is bigger and further away like dandelo suggested.
Thanks for your interest,
Best Regards,
Jan