search

menyifang / MIMO

Official implementation of "MIMO: Controllable Character Video Synthesis with Spatial Decomposed Modeling"
Apache License 2.0
706 stars 33 forks source link

How do you do pose align (retarget) during inference? #12

Open TZYSJTU opened 1 day ago

TZYSJTU commented 1 day ago

How do you do pose align (retarget)? Since I do not see it is mentioned in the paper and the demos seem cannot change the human body shape such as thin and fat, maybe you have not completed this?

menyifang commented 18 hours ago

It is not included in our inference and demos, but can be easily implemented by extracting smpl parameters from target characters and using the shape params to render the pose representation.