Common material spec: handling of specular exponent

[mlimper]

As discussed here, this PR adapts the spec to clarify handling of different values for the specular exponent.

Just a strawman proposal - it would probably also be a good solution to leave this special case of adapting values from [0,1] completely out. I guess that could facilitate writing correct reader / renderer implementations even more.

[pjcozzi]

@mlimper this sounds like the client has to map [0, 1] to [0, 128]? Why can't this be done at conversion time? That is the glTF way: keep the client simple.

[mlimper]

Why can't this be done at conversion time? That is the glTF way: keep the client simple.

Yes, I would also prefer that!
Actually, that's what I meant:

I guess that could facilitate writing correct reader / renderer implementations even more.

But then we need to adapt the current spec and remove the remark about handling different ranges differently (the passage about the Blinn-Torrance-Sparrow model).

[pjcozzi]

That is all good with me.

[mlimper]

Ok, thanks - I updated that.

Now, the PR removes the mentioned part of the spec about different interpretations of shininess.
Instead, an implementation note is added, clarifying the interpretation and illustrating typical usage by example (could maybe be more concise, I don't know).

[pjcozzi]

OK with me. @tparisi are you good with this?