>>4452105
>the RGB that humans can detect is fairly broadband, so we'll look at two things and call them "red" but they may not have much spectral overlap
Human perception has almost nothing to do with digitizing film. Displaying it sure, but digitizing is a simple matter of counting molecules of dye per unit area. There are 3 different dyes, so you get 3 different counts.
>LEDs only come in a handful of specific wavelengths, based on their construction
LEDs are available in a staggering variety of wavelengths, in reels of a thousand or more for pennies apiece. It would absolutely be possible to make a light source with LEDs every 10-20nm and get 40+ channels of color. But there are just 3 dyes so most of those would be redundant.