>>4447560
interesting test, I did a similar but much more limited test with my mk1
it showed similarly poor results at 100 pushed, but imo "better" (more invariant) results after that (picrel is a few of the test shots, should really redo it with a better test subject)
>(red + blue = magenta)
in theory the magenta is a feature of any bayer sensor
maybe this is something everyone else already knows, but I figured it out recently writing a digital camera sim for work
you've got more green pixels so assuming SNR is the same for all pixels, you'll have less green noise once you debayer
not half as much, but I did end up with exactly the same magenta tint I get from my K-1
it was cool