One clear insight from those plots is that my earlier idea of using ISO400 which has a lower gain and therefore greater dynamic range wasn't doing me any favours with respect to noise (even though I did see much better star colour). eg this image of Corona Australis from two weeks ago:
http://www.iceinspace.com.au/forum/a...e.php?a=137559
I think these results suggest that a better approach for me to enhance star colour would be to take a separate set of short subs at ISO1600 (say 60s) and then do a HDR combine in Pixinsight.