I am trying to get the monocular render from the smplmesh output and I am struggling a bit.
I added this code to the 2bvh.py file to get the camera to the "right" place, and save the blender file to use the output in blender.
However, I am currently doing some manual tinkering with the camera with set values, that work with this input and haven't yet tested any other video. The closest I got the SMPL model to line up with the video was by:
Using the root as the camera position
.759 to the tail of the root bone, to take the camera down a bit (line 261 in the above code)
Setting the camera to the iPhone's front facing camera focal length
The model is still a bit off the video.
I know the approach I am taking is not robust, and asking for help to see if there are other ways to get consistent camera and position.
Hi,
I am trying to get the monocular render from the smplmesh output and I am struggling a bit.
I added this code to the 2bvh.py file to get the camera to the "right" place, and save the blender file to use the output in blender.
However, I am currently doing some manual tinkering with the camera with set values, that work with this input and haven't yet tested any other video. The closest I got the SMPL model to line up with the video was by:
The model is still a bit off the video.
I know the approach I am taking is not robust, and asking for help to see if there are other ways to get consistent camera and position.
Thanks.