This is an implementation of Tensorflow's pix2pix as a Cog model. Cog packages machine learning models as standard containers.
pix2pix is a general purpose conditional adversartial network for image-to-image translation prediction.
Here are some example outputs from a model trained on a 455 pair dataset of Albedo>Normal cc0 textures:
Models were trained using CC0 data from ambientCG, PolyHaven, and Pixar RenderMan.
Individual models are avaliable on replicate, or you can try out the Unity based AI Material Designer Tool on Monaverse