Welcome for Computer Graphics.

Today we will look into scene graphs and also do a little bit of modeling.

So essentially the idea is that up to now we were always looking at the lower part or

later part of the rendering pipeline.

That is we were looking at how can we rasterize single triangles, how can we transform these,

how can we do shading on these triangles, how can we apply textures.

But we didn't really look at the top above that.

So where do the triangles come from and how is this entire scene content, how is that

generated.

And this is what we will look at today.

So you see if we, I already mentioned a few times that we can organize the entire rendering

process in a pipeline that has various stages.

And we were always looking at these stages now and in the beginning there's a 3D scene.

That is then transformed somehow into triangles and from the triangles on we know how to handle

them.

But yeah, today we will look at this stage here and this stage is a little bit at the

interface between CPU and GPU.

Normally in the beginning we have a scene that is stored on the CPU.

We know the objects in our scene, maybe how they are moving and all that stuff.

And then at some point we have to transfer the data to the GPU and then do the rendering

on the GPU.

And yeah, we will now look at this interface here.

Now where does the 3D scene, how is that modeled and how is it handled and how do we get the

data prepared for rendering?

And for this we need a number of scene objects and if you remember the last lectures for

instance in the beginning we were looking at very simple objects like cubes or single

triangles even only.

And of course that's the normal way to start with.

So we have very simple objects such as spheres, cones, cylinders that can be described very

easily and for which it's also rather simple to transform them to a triangle mesh or a

simular, so for a cube we can simply enumerate the vertices and enumerate how the triangles

finally form the cube.

And the same is true for spheres, cones and cylinders and so on.

But yeah, that would not be a sufficient basis to make a movie for instance or a computer

game.

So in these scenarios we need real modeling tools, programs that allow us to generate

nicely looking characters that in the end are only triangles but at the beginning of

course they are not triangles, more than triangles.

No movie is created by starting with one million triangles and then an artist takes the single

vertices of the triangles and moves them to the right position, that's not the way how

it works.

And this is what we will look into a little bit.

How can we model such nice looking shapes and then convert them to triangle, to triangles.

Another possibility is to use scanned objects, that means take real world objects and then

run any method to capture that 3D shape of the object.

So for instance that bunny that we've also used in several of the demos and that you

probably seen quite often already now is a real world object that has been scanned with

yeah, there are a number of methods available to do that and then this gives a rather fine

triangle mesh and this we can also use for rendering.

And I mean it's a little bit a variant of these objects designed by a modeling tool.