Have you ever wondered how animation artists create near-to-human faces in animations? Think about monsters or caricatures. Well, maybe we don’t know the exact answer, but now we’ve an idea, and you can easily try it by yourself!
Take a 3D Morphable Model #3DMM), like the Basel Face Model (https://lnkd.in/dUKXiDU) that we’re currently using in a very cool project on 3D face reconstruction from #RGBD cameras on the #fashion industry.
3DMM are used on all kind of shapes, not just faces, and different industries (e.g. #medical, #biometrics or #creativemedia, just to name a few).
According to 3DMM, any shape is represented by giving proper values to a bunch of PCA coefficients that apply changes on top of a mean shape. Cool, but what about monster faces? Ok, coefficients are expected to fall within a normal standard distribution (sigma=1). What happens if you push values a bit beyond that limit? Here you go, check the image down here! First row is within sigma=1, second row is a bit further out.
And what about the 3D monster version of your face? Well, just push a bit further the values of your 3DMM coefficients: but you need your coefficients first: there’s plenty of options out there, or join Deep Vision Consulting and you’ll learn how to do it by yourself!