Brain Dump

Scene Graph

Tags
computer-graphics

A [see page 20, graph] representing objects in a 3d-scene.

Leaves should be specific shapes plopped at the origin, then as we go up the graph we [see page 13, apply transformations] at each node to the scene that's being incrementally built, with each transformation being represented as explicit nodes in the scene graph. This has the affect of applying the same transformation to each child of that node in such a way that the transformation appears to be hinged at that point.

Consider a [see page 7, [hand](COM3503-w02-hirearchy-struct+scene-graphs)], each rotation for a tip is done relative to its joint at the middle of the finger and each rotation for the middle of the finger is done relative to its fingers base.

It's quite common to hide joints by placing another shape between them in the scene graph, this is because as they move the two connected parts overlap and interfere and this isn't visually pleasing so having a stop gap is commonly used.

See a [see page 10, step-by-step] conversion from a single sphere into a tree with a stem and two branches.

You can [see page 19, visualize] transformations at different nodes in a scene graph.

Skinning

Commonly we'll be constructing scene graphs with rigid bodies: rectangles, spheres, solid objects. An alternative approach involves just defining a [see page 25, skeleton]: Joints, and the lines between them. We can overlay a model (skin) on top of the skeleton to make it appear as a rigid body. We animate the skeleton (often through motion capture) and have the skin then deform as the skeleton moves to make it appear as animated.

Skin points can be attached to two bones (weighted between them) to better [see page 28, blend] as the joint is moved. We can also have muscle controllers to cause a bulge when the skeleton flexes.

Links to this note