A college student's thesis work has made strides in machine-original art generation — at least in artistic styles that emphasize edge definition.
The student created a data set of over 2,000 traditional Chinese landscape paintings from museum collections to train the model.
Synched reports, "Princeton undergrad student Alice Xue has designed a GAN framework for Chinese landscape painting generation that is so effective most humans can’t distinguish its works from the real thing. ...The proposed framework, Sketch-And-Paint GAN (SAPGAN), is the first end-to-end model for Chinese landscape painting generation without conditional input. The 242 participants in a visual Turing test identified SAPGAN paintings as human artworks with a frequency significantly higher than paintings from baseline GANs."
The works created by SAGPAN were mistaken by the human participants as human art as often as 55 percent.
Read the paper End-to-End Chinese Landscape Painting Creation Using Generative Adversarial Networks on arXiv.