So welcome to all of you for our computer graphics lecture.

So last Tuesday, we were speaking about lighting.

And yeah, I didn't finish in time a last lighting model,

which is called the Torrance-Sparrow lighting model.

And yeah, as I said several times,

the Fong lighting model that is very widely used

has severe limitations.

Things always look a bit like plastic.

It's from a physical point of view.

It doesn't have any justification.

And the Torrance-Sparrow light model now has similarities,

but it's also more physically based.

And the idea is that we consider that what

we model is a kind of macroscopic surface.

And if we would take a microscope

and have a closer look at that surface,

or also at normal surfaces like this blackboard here,

for instance, that has some roughness.

It has some microstructure.

And this microstructure is very important for the visual

appearance of that model.

And if we take a closer look, then we

see that this microstructure, for instance,

looks like, in that example, like a little bit bumpy

surface.

Of course, there can be more complicated setups,

but that would be an example.

And now the idea of that model is

that light now is reflected by this rough surface,

and it's perfectly reflected.

So there's a perfect reflection, like a mirror.

But since the surface has some roughness,

we don't perceive it as perfect reflection.

So this roughness of the surface results

in a non-perfect reflection.

And now in order to model that and to describe that,

now three different things are considered in that model.

And the most important one, of course,

is that the reflection depends on how rough the surface is.

And now measuring roughness or describing roughness

is a bit difficult. And what they do in that model

is that they describe the distribution of normals

over that surface.

So they say, or it is said that if this

is the main normal, for instance,

then a particular surface has some variation

of the normal direction.

So for each normal direction, a probability is given.

So if we only have 2D, for instance,

then the varying normals could be described by an angle,