Additional materials
2D or not 2D: How Does the Dimensionality of Gesture Representation Affect 3D Co-speech Gesture Generation ?
Accepted as an extended abstract at IVA 2024
Téo Guichoux, Laure Soulier, Nicolas Obin, Catherine Pelachaud
Accepted as an extended abstract at IVA 2024
Téo Guichoux, Laure Soulier, Nicolas Obin, Catherine Pelachaud
This site contains additional videos for our paper submitted at ACM IVA 2024.
The videos correspond to examples of pair-wise comparison of animations generated by the different models we consider in our study (Trimodal or DiffGesture). A description of the order (3D native first or 2D+VideoPose3D first) is given above each video.
1 ) Trimodal 2D + VideoPose3D (left) vs Trimodal 3D (right) - sentence 27
2 ) DiffGesture 2D + VideoPose3D (left) vs DiffGesture 3D (right) - sentence 3
3 )DiffGesture 2D + VideoPose3D (left) vs DiffGesture 3D (right) - sentence 11
4) Trimodal 3D (left) vs Trimodal 2D + VideoPose3D (right) - sentence 15
5) Trimodal 2D + VideoPose3D (left) vs Trimodal 3D (right) - sentence 2
6) DiffGesture 3D (left) vs DiffGesture 2D + VideoPose3D (right) - sentence 22