ACE: Adversarial Correspondence Embedding for Cross Morphology Motion Retargeting from Human to Nonhuman Characters
easypapersniper.github.io/projects/ACE...
ACE: Adversarial Correspondence Embedding for Cross Morphology Motion Retargeting from Human to Nonhuman Characters
easypapersniper.github.io/projects/ACE...
iykyk
Animating non-humanoid characters with human motion data
https://www.youtube.com/watch?v=h_HAVnDCfkw
Semantic Deformation Transfer
https://people.csail.mit.edu/drdaniel/research/baran-2009-sdt.pdf
Learning to Correspond Dynamical Systems
https://arxiv.org/abs/1912.03015
Creature Features: Online motion puppetry for non-human characters
https://www.youtube.com/watch?v=1AVvHXInVoA
C-3PO: Cyclic-Three-Phase Optimization for Human-Robot Motion Retargeting based on Reinforcement Learning
https://arxiv.org/abs/1909.11303
Kinematic Motion Retargeting via Neural Latent Optimization for Learning Sign Language
https://arxiv.org/abs/2103.08882
Skeleton-Aware Networks for Deep Motion Retargeting
https://deepmotionediting.github.io/retargeting
Human Motion Control of Quadrupedal Robots using Deep Reinforcement Learning
https://sites.google.com/view/humanconquad
who wants to work on this
big challenge will be training generative models to produce 3D models, rigs, etc. for these avatars, as well as motion retargeting from humanoid inputs (like full-body tracking) to non-humanoid skeletons
homuncular flexibility -> a future with a wide variety of non-humanoid vtuber/vr avatars, like lanier's lobster or stelarc's walking head
- https://stanfordvr.com/mm/2016/11/won-etsbs-homuncular-flexibility.pdf
- https://www.youtube.com/watch?v=YyuBuI1Vtmk