Anatomy-Based Human Neck Modeling

    Although significant achievements have been made in the research of human body and facial animation, many difficult issues in human modeling remain unsolved, among which neck modeling has been overly underemphasized. Movement of the neck in animation is often simplified and remains stylish. The difficulty mainly comes from immense complexity of cervical anatomical structure, the elaborate system of joints and muscles for control, and complex interactions between the variety of components (cervical vertebrae, muscles, soft tissues, and skin). The construction of an accurate physical neck model is, however, essential to expressive character animation since the neck generates flexible head movements that are fine tuned to the abilities of human in handling numerous tasks. Indeed, the neck plays a vital role in providing stability of the head in a given posture, and in consequence controls the entire eye-head-neck system.
    Since the neck movement is produced fundamentally through the combined articulations of both rigid bones and soft tissues, an accurate model that has a biomechanical basis is required to capture the key characteristics of neck motion. And due to the complex structure of the human neck, only complex dynamic models based on human anatomy can reflect adequately the dynamic properties of the neck. Physically-based modeling of the anatomical head-neck complex is an appropriate way to explore dynamic neck function. Its ability to simulate and animate biomechanical neck motion that results from neuro-muscular excitation and is very difficult to be described heuristically makes such a method superior to procedural simulation of the motion. Although data-driven methods offer alternative approaches to anatomical models for creating realistic motions, the motion synthesis is limited in its expressive power by the number and variety of the examples in the data set. Physically-based models, in contrast, allow direct control over the subtleties of the movement using a single representation and can generate any new motion without additional shape digitization. A range of dynamically simulated behaviors could then be used as an interpolation mechanism to fill in coarsely keyframed example data, greatly augmenting data-driven motion synthesis systems. All the more so because the long-term goal of human animation is to create lifelike characters that are constructed with an anatomical structure of the kind we describe in this paper and able to synthesize a wide range of human motions.
    A synthetic neck model that is anatomically and physically valid can help to explore activations of various constituents of the neck in loading conditions. Indeed, there is a pressing demand for the anatomical model that can replace mechanical dummies in order to gain insight into the damage mechanisms of whiplash injury during the car accident. The model with dynamic function may be also of interest in a clinical setting to study the stability and conditions of the cervical spine of patients with spinal cord pathologies. The understanding of basic disorder mechanisms of the neck, however, is still limited and many studies are ongoing experimentally. We hope our model is one step towards a comprehensive biomechanically accurate model that eventually allows experts from different disciplines to investigate hypothetical theories based on their domain expertise.
    Our goal is to construct a 3D dynamic human neck model that features the structure and functions of the archetype in the real world for computer animation. We develop an anatomically consistent model that incorporates the dynamics of different components of the head/neck system. Our neck model with mid-sagittal symmetry consists of geometrically accurate representations of the rigid bodies of the skeleton (head and cervical vertebrae), as well as deformable muscles and soft tissues (ligaments, intervetebral discs, and facet joints). With the incorporated geometrical inertia characteristics, the skeleton is modeled as an articulated multibody system. The interconnecting soft tissues are modeled anatomically in detail. Attached accurately to skeletal components, these deformable structures are tuned to emulate the relevant physiological and biomechanical properties to produce passive resistance. Contractile cervical muscles in different muscle layers are grouped to actuate the skeletal system for simulating four main motions of the head/neck complex. Employing Hill's muscle model, our system performs forward simulation from muscle activation levels based on anatomical data and mechanical laws. Neck movement is governed by the equations of motion which are integrated iteratively in time. To model deformation of the muscles and determine their influence on the appearance of the skin layer, we assign actual geometric shapes to three pairs of major superficial cervical muscles. Our approach makes use of a set of muscle control curves which describe the shape of geometric muscles. Given the position of the skeletal bodies during neck motion, muscle control curves are updated automatically to control the deformation of geometric muscles. The resulting model with anatomical fidelity can generate physically realistic neck motions and skin deformations.