MF3D Release 2

Under Development!

Despite numerous requets for this planned development of the MF3D avatar, the project is currently suspended due to a lack of technical expertise or the right people to outsource it to. If you think you or somebody you know can help with the expression transfer process needed to adapt the model then get in touch.

What’s in MF3D R2?

A major limitation of the method used to produce the first stimulus set MF3D R1 was that facial expression and facial identity could not be co-varied simultaneously. Consequently, MF3D R1 contained an expression subset in which facial identity was held constant (identity M02) and a separate identity subset in which expression was held constant (neutral expression). Using a computational operation known as ‘expression cloning’ (or more generally ‘shape deformation transfer’), we are now able to transfer facial expressions across identities to generate a novel stimulus set of CGI macaque faces with unparalleled parametric control.

MF3D R2 Subsets

The static stimuli of MF3D R2 will extend those of R1 by covarying the following variables:

  • expression types

  • expression intensities

  • head orientations

  • identity dimensions

  • identity distinctiveness levels

  • lighting directions

The animated component of the MF3D R2 stimulus set will replicate the animation subset of MF3D R1, but includes a variety of facial identities of the avatar.