CS 425 Lecture 4

Jan Allbeck

Outline Navigation? Animate the characters Load and place objects Glue/sticky objects/constraints/carrying stuff

Different types: position, orientation, offsets, cameras, limbs

Containers and locations Actions: PickUp Object, Drop Object, Move Picking Assignments

Navigation How did it go? Issues?

Make sure that your code is in the repository. I need to make sure everyone is making good

progress. I need to look it over and see how good/bad it

is. I’ll try to give some feedback.

Animating Articulated Figures

Don’t have time to give a full course, but small introduction to Scene Graphs…

Grouping Transformations in a Scene Graph Motivation:

Often an object has a part or sub-structure.

Repeated instances of an object need not be separately constructed.

Organize transformations. Organize geometry (shapes).

http://en.wikipedia.org/wiki/Scene_graph http://www.openscenegraph.org/projects/osg

Grouping Transformations in a Scene Graph

Create a directed tree of transformations: Geometry at terminal nodes. Non-terminal nodes become transformed

instances. The same geometry primitive be referenced by

leaf nodes and transformed several times resulting in a scene graph.

Pictorially…

Scene Graph ElementsGeneric scene graph node:

Each child points to another (lower) scene tree node or to stored geometry.

TransformationMatrix [.]

Child [vector oftree or geometry

pointers]

Geometry-1

Geometry-2

Geometry-3

Geometry-4

…

Geometry primitives

Sample Scene GraphShown is a 2D shape built from simple geometric parts in a

scene graph of transformations.

G1: (0,0) to (1,0) G2: Unit square

G3:Unit circle: center (0,0); radius = 1

(1,1)

(0,0)

T3> >

T5> >

T4G3

I>

T1G1

T2G2

T1 = Rotate by 90°

x

y

(0,1)T3

> >

T5> >

T4G3

I>

T1G1

T2G2

T2 = Translate by (-0.5, 0)

x

y

T3> >

T5> >

T4G3

I>

T1G1

T2G2

T3 = Scale by 0.5 in both x and y

x

y

T3> >

T5> >

T4G3

I>

T1G1

T2G2

T4 = Scale by 0.15 in x and y THEN Translate by (0, 0.25)

x

y

T3> >

T5> >

T4G3

I>

T1G1

T2G2

x

y

T5 = Translate by (-0.6, 0.4)

x

y

Think of this as a new object (group) instance that can itself be transformed.

T3> >

T5> >

T4G3

I>

T1G1

T2G2

“Executing” the Scene Graph:INORDER Traversal of the Transformation Tree

Traverse (root.child, I);Traverse (N, T):

T= T•N.transformation; //push (multiply) transformation on right

for (i=0; i<N.child()–1; i++) If N.child[i] points to geometry

then apply T to N.child[i] and draw it else Traverse(N.child[i], T)

Recursion works for you: Go up implicit POP transformation from right: [.] T [.]

For our example, we just created all the transformation matrices.

Watch ...

Scene Graph Example

Then transformation of G1 line segment is this matrix product:

x

y

IT3

> >

T5> >

T4G3

I>

T1G1

T2G2

T5 T3 T1

I

Scene Graph Example

Then transformation of G2 square is this matrix product:

(+ two more …)

x

y

IT3

> >

T5> >

T4G3

I>

T1G1

T2G2

T5 T3 T2

Scene Graph Example

Then transformation of G3 circle is this matrix product:

(New radius = 0.15)

x

y

IT3

> >

T5> >

T4G3

I>

T1G1

T2G2

T5 T4

I

Scene Graph Example

The transformation stack is empty so we are done.

x

y

T3> >

T5> >

T4G3

I>

T1G1

T2G2

Scene Graph with “Internal” Geometry

T3> > G2

T5> >

T4G3

I>

T1G1

T2G2

• We added a third child to the node with T3, pointing to geometry G2 (the unit square). The result is that this instance of G2 is only transformed by (T5 T3): scaled by 0.5 in x and y and then translated by (-0.6,0.4).• This feature can be used to create articulated parts (such as character arms and legs), or objects such as tables with supported (sub-)objects.

x

y

Articulated figures in DarkGDK Everything is done for you…

dbLoadObject: already using dbAppendObject: add more animations dbTotalObjectFrames: how many frame so far dbPlayObject dbLoopObject dbStopObject dbSetObjectSpeed

Should scale to other movement

Do not go in the loop

Go to the code…

Lecture 4.0

Assignment 1

Create classes for loading, storing, replaying, and stopping actions. Keep in mind that some actions loop and some do not. Load all actions for all characters.

Loading and Placing Objects

Loading and Placing Objects Note: characters/agents/NPCs vs. objects In spite of documentation, they must be .x

files In our game, weapons and possibly furniture Already have some examples

Go to the code…

Lecture 4.1

Assignment 2

Add additional symbols (numbers?) for objects to the environment set up (at least 3 types). Consider how to represent these objects. Add another class for non-user, non-NPC objects?C:\Program Files\The Game Creators\Dark GDK\Media\Dark Matter Models\Weapons

Object Constraints (sticky objects) Glue Defines transformational relationships between

objects (Scene Graphs) Position and/or Orientation

All dimensions Just some dimensions

Can include limits Can have offsets Can be turned on and off Limbs Show and hide objects (e.g. PickUp) Sticky cameras

Go to the code…

Lecture 4.2

More about objects Locations and contents can be helpful

Collision detection Picking AI

Types too Actions for our game

PickUp Object Drop Object Face or look at or watch Follow/walk with?

Assignment 3

Write the necessary code/classes to load objects and enable gluing/constraining any object to any other object. Think in terms of the needs of our game and games actions.

Assignment 4

Start building a camera class that takes advantage of the built in camera functions and allows for different perspectives and inputs. Allow for multiple cameras and switching between them. Allow user camera interaction. Allow for first person perspective. Enable the camera to follow a selected character. Allow for an overhead camera. Camera collisions on/off.

Picking Using a mouse to choose an object in the

scene. How does the 2D position relate to your 3D

environment? Involves:

Ray intersection Screen vs. world coordinates Camera transforms Depth buffers

Geometry for Ray-Polygon Intersection Test

Converts NDC screen coordinate to world point to determine 3D ray from view center of projection.

(Thanks to Andrew Glassner) Geometric Situation: Know:

Eye world coordinates: E Viewing vector direction: C (implies viewing

distance is |C| ) Up vector: U Field of view half angles: θ and φ

Output is a screen midpoint and two vectors used to find ray in world coordinate space.

Simple Viewing Geometry

UB

A

V

HM

(0,0)

(1,1)

θ

φ

EA || HB || VScreen coordinates are NDC

P=(sx,sy)

C

Algorithm for Needed Vectors and P A C U (be sure this is well defined) B A C (B is now in the correct plane) M E C (now have midpoint of screen) H (A |C| tan θ) / |A| (rescale A to H) V (B |C| tan φ) / |B| (rescale B to V)

Given the origin as shown, any point in NDC is (sx, sy) [0 sx 1; 0 sy 1)]

E.g., if frame buffer is 320 pixels wide by 240 pixels high, then (50, 75) would map to (50/329, 75/239).

P M (2sx 1) H (2sy 1) V The ray equation is thus:

R E t(P E) / |P E| (ray direction D is normalized P E)

Go to the code…

Lecture 4.3

Assignment 5

Use Object Picking to help with testing as desired.

Assignments (All Good Object Oriented)1. Create classes for loading, storing, replaying, and stopping

actions. Keep in mind that some actions loop and some do not. Load all actions for all characters.

2. Add additional symbols (numbers?) for objects to the environment set up (at least 3 types). Consider how to represent these objects. Add another class for non-user, non-NPC objects?

3. Write the necessary code/classes to load objects and enable gluing/constraining any object to any other object. Think in terms of the needs of our game and games actions.

4. Start building a camera class that takes advantage of the built in camera functions and allows for different perspectives and inputs. Allow for multiple camera and switching between them. Allow user camera interaction. Allow for first person perspective. Enable the camera to follow a selected character. Allow for an overhead camera. Camera collisions on/off.

5. Use Object Picking to help with testing as desired.

Next? More on interfaces Multiple windows NPC AI Animating objects

Drop object Read in environments Need good character models Clean up animations Exchange code to check interface and find

problems.