This tool uses a series of python scripts to turn a single image of a person into a rigged 3d Model usable by mixamo skeletons.
Using the python repo pifuhd, each generation takes approximately 5 minutes, including mixamo upload and auto rig along with texutre map baking on my semi-humble rig.
The user adjustments are on the orthographic camera position to match the image to the generated model, and then run the mixamo auto-rig on the processed 3d model.
It then runs a script that downloads all the mixamo animations into a directory which is then parsed by python to isolate a few search queries.
While the results are not perfect, this could provide useful for low detail crowd agents.
