Blackbody curve to floating RGB

Question

I want to set the color of the lamps in my game with a black body curve, but the approximations I can find for it are all in sRGB color space; and I want it in linear floating RGB.

The closest thing i was able to find is this: https://mathematica.stackexchange.com/questions/57389/convert-spectral-distribution-to-rgb-color

But i really don't understand most of the notation in the answer.

Mainly I'm unsure: if i just use planck's law, using the wavelengths of the subpixels in a standard monitor--not that I know what those are--will it get the right answer, or would it need to be an integral with respect to cone response.

Furthermore I'm not sure how to normalize the resulting radiation density from.... whatever units its in, to 1.0 = X nits; as I'm unsure of the conversion between the units...

Answer 1

The Open Shading Language has a complete implementation which is based on this source code from Color Rendering of Spectra by John Walker, according to the comments. I can't vouch for its theoretical correctness but I have used it and it appears to work reasonably well.

Answer 2

The code below is based on the article How to Convert Temperature (K) to RGB: Algorithm and Sample Code with Neil Bartlett modifications from the article Color Temperature Conversion Of Homestar.io.

Bartlett has released a javascript library neilbartlett/color-temperature which computes RGB from kelvin with the color components in the range 0 to 255. The code below is adjusted for RGB components in the range 0.0 to 1.0.

float t = kelvin / 100.0;
float x;
float red;
float green;
float blue;
if (t <= 66)
{
    red = 1.0;
}
else
{
    x = t - 55.0;
    red = clamp(
        1.38030159086 + 0.00044786845 * x - 0.15785750233 * log(x),
        0.0,
        1.0);
}
if (t <= 10)
{
    green = 0.0;
}
else if (t <= 66)
{
    x = t - 2.0;
    green = clamp(
        -0.60884257109 - 0.00174890002 * x + 0.40977318429 * log(x),
        0.0,
        1.0);
}
else
{
    x = t - 50.0;
    green = clamp(
        1.27627220616 + 0.0003115081 * x - 0.11013841706 * log(x),
        0.0,
        1.0);
}
if (t <= 20)
{
    blue = 0.0;
}
else if (t <= 66)
{
    x = t - 10.0;
    blue = clamp(
        -0.99909549742 + 0.00324474355 * x + 0.45364683926 * log(x),
        0.0,
        1.0);
}
else
{
    blue = 1.0;
}

Note: log refers to natural log.