Monocular, One-stage, Regression of Multiple 3D People and their 3D positions & trajectories in camera & global coordinates. ROMP[ICCV21], BEV[CVPR22], TRACE[CVPR2023]
Hi,
When I used romp estimated output poses to visualize using codes(batch_rodrigues, lbs, etc.)it didn't fit with poses -> Euler XYZ(using mathutils) in maya.
Here are the results (the gray mesh is visualized output and the light orange color mesh is from f_smpl 1.0ver from the official page)
Only global rotation is working well..
The difference is more than just body shape/pose blendshape adding.
@whatyougonnado
I thinks this is because we use different gender of SMPL model.
ROMP estimates the parameters of Neutral SMPL model.
Your f_smpl 1.0ver may indicate that it is a female version of SMPL model.
Hi, When I used romp estimated output poses to visualize using codes(batch_rodrigues, lbs, etc.)it didn't fit with poses -> Euler XYZ(using mathutils) in maya.
Here are the results (the gray mesh is visualized output and the light orange color mesh is from f_smpl 1.0ver from the official page) Only global rotation is working well..
The difference is more than just body shape/pose blendshape adding.