Presentation is loading. Please wait.

Presentation is loading. Please wait.

Physics / AI Week 4 GAM 224. Outline Announcements Demos Discussion Break Lecture Physics AI.

Similar presentations


Presentation on theme: "Physics / AI Week 4 GAM 224. Outline Announcements Demos Discussion Break Lecture Physics AI."— Presentation transcript:

1 Physics / AI Week 4 GAM 224

2 Outline Announcements Demos Discussion Break Lecture Physics AI

3 Announcements Games on order now here Game lab move

4 Demos Civilization 3 (Mortega) Total Annihilation (Reedy) Evil Genius (Kaplanskiy) Railroad Tycoon 3 (Johnson-Vinion) Sims 2 (DiConsola)

5 Game physics Not Physics 101 physics what happens to objects in the game world and why usually doesn't include decisions of non-player characters

6 Example: Asteroids What is the physics of Asteroids?

7 Basic fact All physics is simplification complex multi-body physical simulations are too slow reality is chaotic limits to what can be rendered graphically

8 Example: Halo What happens when a grenade explodes? do we simulate the ignition and rapid oxidation of explosives, pressure waves, metal shear and shrapnel trajectories? do we simulate concussion injuries, soft tissue damage, and bone trauma?

9 State The state of the game representation of the unique instant in the game world what the "save game" feature should save In a card game each player's hand the deck In Asteroids?

10 Game physics Physics = the evolution of the game state we want the player to feel as though there is a real world in the game the game state must be complex its evolution must seem natural the player's control over it should seem natural

11 Natural? Games are profoundly unnatural Aliens? Psychic powers? Controlling a civilization over centuries? As in fiction "willing suspension of disbelief" natural within the game world context Game physics may have nothing to do with Newton's physics

12 Scripting When there is a fixed stereotyped response to an action in the game we say it is "scripted" As opposed to "simulated"

13 Example Designer decides what should happen when a grenade explodes x amount of damage to all units within certain radius x/2 damage within a larger radius leaves a certain "stencil" on the floor or wall Simplifying the actual physics

14 Example Locked door If player ties to go through locked door with key in inventory, the door opens and key is used up Consequences Doors cannot be battered down, removed from hinges Locks cannot be picked

15 Advantages of scripting Much, much faster to apply a simple rule than to run a physical simulation Easy to write, understand and modify

16 Disadvantages of scripting Limits player creativity Players will try things that "should" work based on extensive physical intuition Will be disappointed if they don't Game will need many scripts predicting their interactions can be difficult complex debugging problem

17 Simulation Will still be a simplification Represent the quantities of interest represent the forces that act on them create physical laws for the game world evolve the game state according to these laws

18 Benefits of simulation More player options Designer doesn't have to anticipate every way to do something Physical laws reusable Do not have script every object Can build (or buy) generic physics engine

19 Disadvantages of simulation Speed extensive simulation may make the game too slow Memory game state may become much larger Testing difficult to test all possibilities

20 Design decision How much to simulate? Where player creativity is important Where realism is important Where a simple enough model can be built What level of detail is required? depends on the constraints of the game always a computational cost

21 Example Script when player enters room, guards converge and attack Simulation #1 when player takes a step, sound is heard over certain radius if guard is within radius and in room, guard will converge and attack Simulation #2 when player takes a step, volume of sound is calculated based on level of stealth, floor material, etc. sound is propagated through room and attenuated based on room contents guards receive sound signal and if loud enough to reach attention, they will move in the apparent direction of sound Simulation #3 same as #2, but in 3 dimensions, through floors, etc.

22 AI "Physics" of non-player characters simulating the opponent simulating other characters In some games it is the whole point classic strategy Sims

23 Artificial Intelligence Broad field of study 1950s: how to make computers think like people Now how to make computers less stupid

24 Game AI Uses very little from AI research too slow focused on different problems Main issue "planning" how to get characters to act believably

25 Planning A planning problem Current state End state Known operations preconditions postconditions Construct a sequence of operations to transform current state to end state

26 Example Current state player is alive End state player is dead Known operations move around use machete precondition: must be near target postcondition: target is dead

27 Means-ends analysis Work backwards from desired goal to current state Example to use machete, must be near player to get near player, move etc. Problem the world changes all the time

28 Modern planning Much more complex Integration of acting sensing choosing prioritizing etc. Overkill for games

29 Typical game AI tasks Character behaviors usually scripted actions Route finding many games Resource management in strategy games Modeling character relationships in some RPGs, some strategy games

30 Goals for AI Provide player with challenge Make a good game experience May go against "intelligence"

31 Issues Simulate or script? as in physics Opponent should not make obvious errors Opponent should not be totally predictable

32 Example Simulation when player is firing, construct a line-of-sight map from player's position. Find area outside of line-of-site especially if it is large and close Script have predefined "hiding" zones when attacked go to the nearest location In-between have predefined "hiding" zones calculate which is in the line-of-sight go to closest re-evaluate line-of-sight and move

33 AI in design Not too smart Don't set expectations too high Game play is more important Not too dumb Must cover minimum player expectations Compensate Level design Resource advantages

34 Role-playing Demo Ballerini: Tales of Symphonia Bubon: Paper Mario Devlin: Pokemon FireRed Miranda: Knights of the Old Republic Summers: Fable Non-demo Bennett: Tales of Symphonia Dannemiller: Worlds of Warcraft Hahn: Diablo 2 Abero: Knights of the Old Republic Lipsey: Final Fantasy X-2


Download ppt "Physics / AI Week 4 GAM 224. Outline Announcements Demos Discussion Break Lecture Physics AI."

Similar presentations


Ads by Google