Perler Beads to Landscapes

A little experiment that creates realistic looking digital images from perler beads patterns.

Perler Beads to Landscapes is a small experiment that attempts to add a physical aspect to AI-generated GAN images. The idea came after having tried NVIDIA GauGAN, named after post-Impressionist painter Paul Gauguin, which creates photorealistic images from segmentation maps. A crowd-trained replica of GauGAN called SPADE Landscapes was ported to RunwayML by Gene Kogan allowing someone like me to interface it .

I used Processing to have a webcam focus on physical perler beads and did a bit of hacky programming to have the colors match the colors from the segmentation map. See the video above for a walkthrough.

The interaction opens up loads of playful opportunities and is something I hope to explore more in the future.