This jupyter notebook provides a sample interpolation using the GAN from https://github.com/rp2707/coms4995-project
It trains on the Shapenet dataset particularly two models, a chair and a airplane (~17600 3D volumetric objects). Goal is to interpolate between two noises, one giving a artificially generated chair and another one giving a airplane.
The interesting part is the vector algebra you could use to create mixtures of Objects like: Chair/2 + Airplane/2 = mix of both
Requirements:
- Python >= 3.
- Tensorflow
- Scipy
- Jupyter notebook
- numpy
- Matplotlib
the shellscript downloads the pretrained models (no training needed) and the 3dshapenet data. After downloading you can check the jupyter notebook