Case study
Z-BodyX — merging body scans
NECTEC tool for merging body parts captured by several 3D scanners: landmark picking, thin-plate-spline fitting, symmetry, and motion retargeting.
FIG.1Research
- Role
- Developer for NECTEC (2021–2023)
- Stack
- PythonFastAPIthree.jsJupyterDocker
- Result
- Scans from different devices become one body that can be measured and animated.
Problem
Partial scans from different 3D scanners had to be aligned and merged into one usable, riggable body.
What I owned
Developer for NECTEC (2021–2023)
What I built
- Redesigned the landmark UI with raycaster picking and guidelines.
- Integrated the TPS fitting API and glTF/BVH retargeting.
- Wrote body and head/face symmetry processing in Python.
System shape
FastAPI backend serving a three.js front end where users click landmarks on scanned bodies with guidelines. A TPS API fits parts, Python modules handle body and head/face symmetry, and glTF/BVH retargeting animates the result. Auto-rigging was evaluated and packaged in Docker.
What changed
Scans from different devices become one body that can be measured and animated.