A COMPARISON STUDY BETWEEN RESULTS OF 3D VIRTUAL FACIAL ANIMATION METHODS: SKELETON, BLENDSHAPE, AUDIO-DRIVEN TECHNIQUE, AND VISION-BASED CAPTURE

 In this paper, the authors explore different approaches to animating 3D facial emotions, some of which use manual keyframe facial animation and some of which use machine learning. To compare approaches the authors conducted an experiment consisting of side-by-side comparisons of animation clips generated by skeleton, blendshape, audio-driven, and vision-based capture techniques.

Ninety-five participants viewed twenty face animation clips of characters expressing five distinct emotions (anger, sadness, happiness, fear, neutral), which were created using four different facial animation techniques. After viewing each clip, the participants were asked to score the naturalness on a 5-point Likert scale and to identify the emotions that the characters appeared to be conveying.

Although the happy emotion clips differed slightly in the naturalness ratings, the naturalness scores of happy emotions produced by the four methods tended to be consistent. The naturalness ratings of the fear emotion created with skeletal animation were higher than other methods.Recognition of sad and neutral were very low for all methods as compared to other emotions. Findings also showed that a few people participants were able to identify the clips that were machine generated rather than created by a human artist.The means, boxplots and HSD revealed that the skeleton approach had significantly higher ratings for naturalness and higher recognition rate than the other methods.