The embedded video shows a frame-by-frame translation of gameplay (source) from Mario into the corresponding Voronoi diagrams. This was part of an attempt to implement Kopf and Lischinski's SIGGRAPH'11 paper 'Depixelizing Pixel Art' in a team of two. Code available on GitHub.
PS: Notice the reduced blockiness (since the algorithm infers shape) when Mario is on top of Yoshi. The video is a bit blurry due to YouTube’s compression, but it’s still an improvement over the original!