RE: Character Objects: Properties

From: Bullard, Claude L (Len) (clbullar_at_ingr.com)
Date: 19 June 1998



I think that is somewhat how Crawford's system works. OTOH, if we start clean and define characters as a class hierarchy, we could do a clean room implementation and control characters from Java and Javascript.

Two levels of states:

  1. States of characters within worlds (eg, the internal states of the character) and their transitions.
  2. States of worlds (precondition/postcondition processing) that change when a new world is loaded.

The user has to negotiate the character objects in the scene to get the world to transition to the next world. The user may or may not know what is in the next world. The author won't either: voila, emergence.

We need the character object hierarchy to simplify building new characters, to make it possible to add characters, for a characters properties/methods to be exposed (IOW, all the usual reasons for modeling objects as classes).

Question to Michael: is a role a property of a relationship object?

To do this, we need to think simpler in the beginning so we can use inheritance. For example, "walk/notWalk", 'see/notSee", "talk/notTalk", and so on. At the top levels we should define the most basic behaviors and properties of characters. Then move on to the more complex ones such as roles and emotional states. Eventually one wants to have a property bag that can be passed between worlds.

Not a good programmer here, so folks with more chops, feel free to demolish at will.

Len Bullard
Intergraph Public Safety
clbullar_at_ingr.com

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

> -----Original Message-----
> From: John D. DeCuir [SMTP:decuirjd_at_yahoo.com]
> Sent: Friday, June 19, 1998 10:21 AM
> To: vrml-lit-list_at_kith.org
> Subject: Re: Character Objects: Properties
>
> Here's an interesting idea. I think you all know about states and
> state diagrams. Basically you can have a few states (mad, happy, sad,
> etc.) and some events which spark state transitions between them.
> This idea has been around a long time in CS (especially CS theory).
>
> What if we model our characters as a state machine, but the
> player/viewer's job is to navigate in this state machine to an
> appropriate goal? This is just like wandering around in a maze -- in
> this way the "maze" metaphor from interactive fiction (Zork, etc.)
> makes the transition from space to psychology. Combine the two and
> you have a very complex interweaving of character manipulation as well
> as spatial manipulation. Has this type of thing been done before?
> -John
>



This archive was generated by hypermail pre-2.1.8 : 28 November 2004 CST