Unowns GAN generator, nothing more, nothing less...
Generated ones:
The construction of the GAN is done in two steps (progresive training):
First stage:
Second stage:
It does repeat many of the original shapes, but it is normal considering the few examples we have provided it with. Some new forms come up eventually.
Presentación:
https://docs.google.com/presentation/d/1O7tvokOFY0WprF25QzxtgTfFmqiyF-eJRt6dczwxM_E/edit?usp=sharing
Notebook:
https://colab.research.google.com/drive/1SLpcNh47ojG48jRX-IkJtd1uzTiM-j2x?usp=sharing