Home

Welcome to the helsynky_GAN wiki!

I'd just like to reiterate: you can find this code on google colab: https://colab.research.google.com/drive/1ObYYn649hcB6Pmntl2H3o2n3Ofaq_b5f

Also, just to have everything in one place (and because it's quite difficult to read in one line), here's the main comment at the top of the .py file:

"This code served as my introduction to creating neural art, and can hopefully be yours, too. I am heavily indebted to Simeon Platonov and his dcgan-facegenerator, which can be found here: https://github.com/platonovsimeon/dcgan-facegenerator/blob/master/face_generator.py. Specifically, I made use of the "train" functionality, tweaking and imporving the weights and layers and redesigning procedural image saving to return a single, high-res image every 100 epochs. As such, helsynky trains a new net on the dataset each time you use it, which I find works reasonably well given the aim of creating accessible neural art. I collated the data set from https://www.crwflags.com/fotw/flags/"

You can find the training data (jpeg_flags) on kaggle: https://www.kaggle.com/nicg6645/jpeg-flags