Inverting the new_joint_vecs?

[andrewnc]

This is great work! I have a quick question. After I have processed the motions as described, I have three folders. joint new_joints new_joint_vecs. When training a generative model here, as described in your paper, you would use the full 623 dimensional vector in new_joint_vecs

If you wanted to extract the rotations in new_joint_vecs to apply to an fbx - how would you do this?

I mean to say, it seems the joint order has changed, there is no Jaw joint, and directly removing the continuous 6d rotations and transforming them into quaternions doesn't yield the desired effect.

I'm curious if you have insights here

[shunlinlu]

Hi, @andrewnc

We actually do not use rotation in our representation from our experiments in the appendix. And we followed Humanml3D format to visualize the motion using position. If you are seeking to visualize the motion in software like Blender, you may try this. And I also visualize the position with rotation after FK in my script, and it seems good.

Shunlin

[andrewnc]

Thank you for the response. To make sure I understand correctly, I would take the 623 dimensional data, extract the XYZ positions for each joint, and run something like https://github.com/IDEA-Research/HumanTOMATO/blob/main/src/tomato_represenation/common/skeleton.py#L103 to recover the rotations for each joint?

[bob35buaa]

Hi, @andrewnc

We actually do not use rotation in our representation from our experiments in the appendix. And we followed Humanml3D format to visualize the motion using position. If you are seeking to visualize the motion in software like Blender, you may try this. And I also visualize the position with rotation after FK in my script, and it seems good.

Shunlin
I want to convert tomato representation to SMPL-X format for visualization as shown in Figure 4 and Figure 8 in your paper. Can you provide your scripts for this? Thanks!