brightness = 1 / (fquadratic * dist2 + flinear * dist + fconstant)
where dist is the normalized distance from the object surface to the light; dist = 0 is right at the light, dist = 1 is at the light's max range.
The standard lighting caps maximum brightness at 1.0, which is physically incorrect. If you draw a graph of this you will see a peak near the light. At dist = 0, the brightness resolves to 1 / fconstant , which is the peak brightness. Reducing fconstant to 0.2 will increase the peak brightness to 5.0, for example. The other two parameters control the gradient of the falloff, though flinear is currently ignored by the shader for being slower to calculate.
Per-pixel lights are troublesome, requiring a lot of calculation, it would not perform well with more than 8 lights. Most current games are limited to 4 dynamic lights per pixel. Morrowind itself doesn't activate more than 8 lights, but patching its renderer shouldn't be that hard. I can experiment, but most options tank performance.
- Days are much brighter, maybe a bit too much. This could be a matter of personal preference, though, and think can be adjusted using gamma correction.
Also, I have one more question: is it possible (not at this stage, only in theory) to have static objects show in the distant land only under certain conditions (only at night, only with rain, etc.)?
The actual bump map calculations aren't very well documented, so anything with non-standard bump matrices won't match correctly yet. Daytime is brighter because the combination of sunlight and ambient is higher than 1.0. If ambient was properly adjusted, instead of relying on saturating to white, it would match better. It's definitely possible to have different categories (there's already classifiers for distance, grass, etc.), but the more categories the slower things get.
Note that these are not all of the meshes I have observed becoming 'purple'. Two of them are modded (the argonian skin and the bottle), one of them is vanilla content (the bonemold curiass). In the case of the curiass and argonian bodyparts, they display in the default posistion slightly above where their actors are. The bottle displays in the correct posistion but appears untextured.
Not sure what's wrong with these, they look normal. Could you post another mgeXE.log with the new version?
It may be due to the lower res textures. Find the Data files/distantland/statics/textures directory and move it elsewhere temporarily. See if that fixes it. It's probably a driver bug, as I've tested those meshes a lot without issues.