Improve blackbody node implementation

[niklasharrysson]

The approximation used in our current implementation of blackbody is limiting the temperature range to a lower limit of 1667 Kelvin. This clamping makes it impossible to get blackbody color results in the reddish colors.

Perhaps a better approximation can be found that permits a temperature range down to a lower temperature, preferable all the way down to The Draper point, around 800 Kelvin, which is the limit for visible light from a blackbody emitter.

This issue was spawned from this Slack discussion:
https://academysoftwarefdn.slack.com/archives/C0230LWBE2X/p1773095319965469

[zeel2104]

I would like to work on it. Could you assign it to me?

[niklasharrysson]

I would like to work on it. Could you assign it to me?

Thanks @zeel2104!

[niklasharrysson]

Suggestion from Arnon in the Slack thread linked above:

We've found that what works well for us for OSL is to simply use the build-in blackbody() function in conjunction with the build-in luminance() function to "normalize" the result, like so:

function color mx_blackboy(float temperature)
{
    color out = blackbody(temperature);
    return out / luminance(out);
}

So for Mtlx/OSL this would be an easy fix. It would then give us a good reference to test against when improving the GLSL implementation.