@Biology Great explanation! (And I'm still laughing about the username change. 🤣)

On 11/27/2022 at 1:21 PM, Biology said:

it is a bug that requires changes to the approximation for the Mie scattering phase function and the tonemapper

Is this helping to trigger the infamous "magenta" pixel flag?

22 hours ago, blingthinger said:

Is this helping to trigger the infamous "magenta" pixel flag?

Don't think so, no. But if the sun, the glow around the sun or the glow that descends to the horizon becomes magenta while everything else is fine, it likely is caused by Mie scattering. I never had the magenta, so I don't exactly know when and where it occurs.

44 minutes ago, Biology said:

so I don't exactly know when and where it occurs

🙂 How have you missed that drama!?

Ben has been using magenta to flag pixels/fragments that have gone off scale (Inf/nan) in the lighting level dept. I figured that if the Mie scatter shaders are shooting out too much light right now, that maybe it would be triggering this debug behavior. Now, I don't know if that means Inf in the floating-point 16/32 bit sense (or whatever bit size floats they're using), or if it means just greater than the light conservation eqns permit, but it seemed a plausible explanation.

I recall now too that people were seeing it show up at low sun levels after doing a fast time-of-day movement. Given that Mie-friendly particles tend to be in the inner 2km altitude and the sun is shooting primarily through that layer at dawn/dusk, maybe it really is contributing?

Meh...when it comes to fragment shaders, there's a million possibilities. Nor does that explain why the magenta was mostly showing up in 9xx Nvidia cards...

Bit of tweeking to close out the day