Figure 10
From: Lip-Synching Using Speaker-Specific Articulation, Shape and Appearance Models
Texturing the facial mesh with an appearance model for OC. (a) Original images that will be warped to the "neutral" mesh displayed on the right. (b) shape-free images obtained: triangles in white color are not considered in the modeling process because they are not fully visible from the front camera. The left image displays the mean texture together with the "neutral" mesh drawn with blue lines. (c) resynthesis of the facial animation using the shape and appearance models superposed to the original background video.