atlv

alternatively to avoid a division, `(normal.xyz * scale.yzx * scale.zxy).normalized()` (component-wise multiplication)

If any component of scale is zero it will cause a division by zero and result in infinite-length normals, maybe the division-skipping version that uses only multiplications is a better...

No, the multiplication-only version will work fine, i'm talking about only one component being zero. ~~`normalize_or_zero` is unneeded.~~ *edit: if the normal is perpendicular to the axis being flattened along...

Thanks for the patience and the fixes

There's a bigger problem with switching to nlerp i've discovered: Glam's nlerp is not commutative, nor can it easily be made commutative. This is because it is normalizing after interpolating,...

I've realized there actually is a simpler way forward that can be implemented now without changes to glam.

`BlendInput` says `additive` determines "Whether or not to additively blend this input into the final result" which is actually not what its doing. Additive BlendInputs are being applied in the...

I think its important to keep in mind that meshopt is usually not used at runtime, its only depended on for asset preprocessing. I highly doubt anyone's gonna be running...

I dont see why it should be in standard material docs, but I could add it there. There's no specular_occlusion input, the occlusion_texture field only affects diffuse. This is an...

Yeah it would just mean adding another g-buffer texture on webGL, which could hurt performance. I think that can be done in a follow up with its own perf testing...