Incremental Realism

• Forearm deformation
  • one fairly easy improvement is to twist the vertices of the forearm segment as the elbow joint rolls (pronation/supination)
• Shoulder
  • very unrealistic if treated as a simple joint
  • the ranges of values for its DOFs are highly interdependent. Jack addresses this by using a concept they call "joint groups" - built-in algorithmic correlations between the various DOFs.
  • I'd prefer to solve the problem by modeling the shape of the bones and muscles involved.. An intermediate step could be defining solid angles for DOF range instead of a linear range
• Spine/Torso
  • similarly, the torso is very unrealistic if modeled as a rigid segment
  • could be addressed by modelling the spine as a series of vertebral joints, but then a method is needed to calculate where the torso's mass/skin moves as a result of spine position
• Breathing
  • subtle, periodic scaling of the humanoid's body could increase perceived "aliveness"
• Eye movement
• Sound
  • humans are not silent animals
  • all sorts of sounds should be produced from the collision and friction simulation engines, as well as vocal noises, both correlating to facial motion and otherwise (e.g. humming)