I haven't done anything with the appearance of nearby stars, though obviously that needs quite a bit of work too.
another related question to ask is how your stars behave, once the sensitivity is increased from visual to a 200 mm photographic lens, say. In Celestia the automag might be a relevant testing ground for this. Anyway, in this case, the distance to the stars remains large throughout.
As a possible reference, here is a nice photo of alpha Cen and beta Cen, taken with a 200 mm lens by Dr. Noël Cramer, Observatoire de Genève.
We see that the two bright stars really blow up in size with the glare regime being very conspicuous. An uncountable number of background stars is visible with the star size remaining tiny.
Where would the hardware-specific profile enter in your approach? Once I increase the amount of collected light by some device other than the eye, I need to know its spectral sensitivity! A CCD, for example is way more sensitive in the blue regime than normal film.
PS: probing your above pixel formula at varying light levels might provide further evidence about its proper normalization and the correct ratio of core / glare size of bright stars.