(Back to INDEX)
                                                                                                                         (PREVIOUS)
 
 

RESEARCH.

REALISTIC HUMAN FACES.
 

         This is a system which generates 3d geometry from photographs and textures them accordingly from the same photographs.  Animation is handled via 2d morphing techniques, from the same photographs.  What is particularly interesting with this software is in how it generates facial expressions.  Traditional systems use a 'Global Blending' approach to morphing.  (see Fig 1.)

FIG . 1  (Global Blend)

                This approach assigns a single weight or percentage to each contributing expression.  The software used here employs a 'Regional Blending' approach to shape morphing.  (see Fig 2.)  Regional blending allows for a broader set of facial expressions by combining faces in a localised way.  The faces are split in a logical fashion according to how they behave and interact with each other.  Weights are then assigned independently to each region.  The example shows a 'fake smile' by combining the top part of a 'neutral' expression with the bottom part of a 'joy' expression.

FIG . 2  (Regional Blend)

                    Like FAMOUS™, 'Realistic Human Faces' employs a 3d paint style interface to modify blending weights. 2d  strokes are applied directly on the rendering of the current facial blend, which is updated in real-time.  The fraction that is painted has a drop-off determined by the opacity of the brush.
 

                    Here is an example of 'Realistic Human Faces' in action.
                                                              POEM_LARGE.MOV  (19 MB!!!)

(Back to top of page.....)