MiniMonet

Create your own AI Art online!

Wait for the model to load, a “Model Loaded” message will appear when it’s ready

Draw inside the canvas on the left and then press the “Paint” button, your sketch will be converted into a painting. Doesn’t work on mobiles (yet).

Want to see some other examples of AI Art I’ve been working on? Click here!

Loading Model…

Input

Model Output

Ground Truth (Real)

  • Dataset size: 10,075 landscape paintings (256×256 px)
  • Training time: 235 epochs (around 18 hours)
  • Hardware: NVidia RTX 2080Ti
  • Implementation: TensorFlow
  • Web Implementation: TensorFlow.js

V2: The training images were first sharpened twice before tracing the outlines, because the lines in the original paintings were not sharp enough to give a decent level of detail. The sharpened edges were then matched up with their pre-sharpened painting counterparts.

MiniMonet is based on the Pix2Pix Conditional Generative Adversarial Network – see the paper here.

Isola, P., Zhu, J. Y., Zhou, T., & Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1125-1134).