Virtual Human Ontology

Levels of Knowledges

For the creation of this ontology, we have considered that Virtual Human (VH) creation can be represented in different levels of knowledge:

• Geometry: The geometry is the physical visual representation of the Virtual Human and is composed of two parts, primary: body shape, and secondary: accessories, garments, etc.
• Animation: Virtual Human’s animation can be prerecorded as Key Framed or Motion Captured. Also it can be procedural animation where algorithms drive movements, like walking, reaching, taking, etc. In order for an animation to take place, the character should have a skeletal structure. This structure is defined in the Structural Descriptor.
• Behavior and Cognition: algorithms that drive the behavior and cognition of the character considering individual descriptors.

When creating VHs, other abstract concepts may have to be considered, they are:

• Morphology: contains information like: Age, weight, height, gender.
• Individuality: Representation individual descriptors, to specialize animations, motions and behaviors.This individuality is can be represented by personality or emotion models.

Moreover, VHs have extra attributes, like Garments and Smart Objects.

Virtual Human Properties

Based on the described levels of knowledge, the structure of the VHO is as presented in the figure. We make explicit that the Virtual human has Morphology descriptor, Individual descriptor, skeleton, geometry, animation sequences, controllers, and some accessories like Garments and Smart Objects.

There are some concepts that are related with the top level ontologies. This is that the concepts in the VHO are the same as the top level ontologies. For example, geometry in the VHO has its representation in the CSO as a Geometrical representation.

Morphology and Individuality

Particular attributes are very important at the moment of personalizing virtual characters. Some of these descriptors are attached to the shape, such as the Morphological descriptors. Some of this information is very important at the when searching for models.

Individual attributes are more abstract, and they are important for annotating animations with individual traits of personality or emotions. Also they are considered for behavioral algorithms.

Geometry

The Virtual human has a Geometry that is a 3D shape. At the same time the geometry can have structural features associated. In this ontology we consider that VH have an skeletal structure that are composed by different kinds of nodes. The types of nodes we can have are Segments (segmentation of the shape), Landmarks (feature points) and Joints (articulations).

Structures can serve for shape analysis, or for driving animations.

Animation Sequences

Prerecorded animation can be represented as KeyFrame or Motion Capture, and can be encoded with different formats (e.g. MPEG-4, VRML, 3DSMax, etc.). In order that an animation can take place, the character should have a skeletal structure, so animations will require some nodes form the structure.

Animations can have some extra information; for example, Motion Captured animation can be described by the morphology of the actor. In the case that animations are intended to express an emotion, this can be also annotated, in order to find animations that express a particular emotional state.

Animation Controllers

Controllers are algorithms used to produce animation. They require inputs to work and produce outputs, which are usually animation sequences or specific joint values. These controllers have an animation propose, such as Cinematic, Physical, Cognitive, Path Planning, Skinning, Sensorial, Gesture, and so on.

Controllers can be already implemented in a software tool. And they may require other controllers to work, animations or virtual humans with some specific individual or morphological characteristics.