Will it be used in real-time systems or purely for static rendering?
Sydney will be use within a real-time, multi-user CVE based in the CAVE.
What is the intended polygon count?
As high-quality as possible while operating within technical constraints for real-time rendering of two such avatars within the CAVE. 80,000 polygons is the target, but the method I will present is easy to reduce or increase this value.
Will it's primarily be an avatar or an agent?
Sydney will be used as both an avatar and an agent.
Are facial morph-targets required?
Facial expressions are an important part of Sydney's intended use. When designing high-quality avatars in particular, this should be staple.
Is eye-rotation required?
Yes, the eyes must be separate meshes capable of independent rotation. A pupil dilation morph-target would also be desirable.
Will the character be used as a host for motion-capture data?
Possibly. Therefore, we will ensure that Sydney's biped skeleton is compatible with our motion-capture standard.
What is appropriate clothing for your character's intended use? We must also think about character movement: if your character is wearing a skirt for instance, how will she sit down without deformation or cloth simuation?
Sydney will be wearing jeans and tight vest top. The CVE platform is currently not capable of complex cloth deformation, so we will keep it relatively simple.
Other issues which seem relevant for your character and intended use should also be considered at this stage.