Phil there are a few things happening here. Top quality lenses are not designed with HA wavelengths in mind when minimising CA. So in the case of the 300mm F2.8L it is a matter of compromise when focusing as to what end of the spectrum is slightly out of focus. The Canon 300mm F2.8L has one lens element made from a single crystal of Fluorite and two ED elements in main front lens groups. Even then it is not perfect.
The CA near the corners are slight aberrations due to the fact no lens at f/3.5 can be perfect. It can be minimised by using f/5 but then signal goes down and thermal noise rears it ugly head due to the longer exposures needed.
In the case of your 50mm lens CA and aberrations are worse due to the wide field of view. That is why even at 'perfect' focus both red and blue are slightly out of focus.
You are quite correct that careful star reduction at the red or blue or both end of the spectrum can help.
The large images of the mosaic are actually upsized by a factor of 1.6 from native pixel size of 4368X2912 to 7014X4668. This with dithering the mount between exposures when collecting really helps with star profiles as at 300mm FL the stars are undersampled. This only works if the lens has more resolution than the sensor.
Here is the image with adequate star reduction of the blue channel, slight Richardson Lucy enhancement and a boost in saturation and reduced to native pixel size.
Large Image 12MB
http://d1355990.i49.quadrahosting.co..._04/ltsatL.jpg
If I simply processed at native pixel size the stars all end up as lovely squares!
Dithering and upsizing before stacking would really help widefields at 50mm to 100mm even more as long as the lens is of high enough quality.
Bert
