I’ve found some other relatively high CRI (97) white LEDs that I’m going to try - they were quite cheap, so it’s worth the experiment before looking at anything more expensive, I think.
The benefit is I can easily fit these LEDs into the current mechanical and electronic design. However, unless it works absolutely brilliantly, I’ll probably then experiment with a 5x5 matrix of R/G/B LEDs driven by three constant current sources for comparison.
From what I’ve read above which is most ‘illuminating’, I’d expect the R/G/B approach to give better colour separation than high CRI white LEDs, ie saturation, but possibly with worse colour accuracy. In the end though, it’s completely subjective, because I am just converting old family home movies so all that really matters is the family’s opinion.
An added problem is that the films (mainly 1960s) have not been well looked after, and in some cases there is fading. Not only that, in a lot of cases the original exposure was atrocious, and transparency film like Kodachrome or Ektachrome has a really narrow tolerance of under/over exposure, but there’s not much one can do about that.
As a point of interest I do have a 35mm film scanner I still use occasionally for slide scanning. It gives phenomenally good results (on well exposed slides). Must look into what light source that uses.