Hector Martin

Well, you could compute a 3D space of the optimal spectra representing a chromaticity from 100% sRGB blue to white, and then pick the one that matches the brightness of 100%B in sRGB (which will be washed out)

I’m pretty sure sRGB (or any other gamut) doesn’t specify brightness.

Relative to white it does. A reflector producing optimal sRGB blue will be much dimmer relative to a pure white 100% reflector than a #00f pixel is relative to a #fff pixel.

Your second sentence is true, but I don’t think that’s actually defined by the color space, since chromaticity is supposed to be independent from luminance. It’s just that displays that use primary-colored subpixels can produce more light when all the primaries are active…

…but the ratio depends on the display, e.g. some displays have white subpixels.

DIsplay technology is irrelevant. The definition of the sRGB colorspace doesn't care how you implement it. White subpixels are a hack to get better brightness at the expense of color purity, and never help you if you're trying to cover the sRGB gamut.

I posit that any display using white subpixels and assuming subpixel independence (no weird global power limitations/interactions or whatever) that covers the sRGB gamut can also cover the sRGB gamut at identical brightness with the white subpixels always off.

ok, even though my previous statement was wrong, *that* isn’t necessarily true, since even the purest blue available is close to pure 470nm light, which is well within the range where M and L are significantly activated in addition to S; therefore…