Still paintings and drawings are endless sources of fascination, not only due to the diversity of styles employed by artists, but also thanks to the way shapes, materials and space are conveyed with an economy of means. Likewise, traditional 2D animations (i.e., where each frame is drawn by hand) convey 3D motion and layout with a few carefully chosen lines, striving for legibility and expressivity rather than realism. However, producing traditional animations requires tremendous skills, time and dedication. We explored two different ways to help produce such animations on computers. The first method starts from a few 2D key drawings, and generates intermediate frames through interpolation techniques. Here the goal is to progressively increase the complexity of moving and deforming 2D drawing parts until they convey 3D motion and layout cues. The second method works with 3D animations rendered in a line art style, and relies on 3D shape deformations and spatio-temporal perturbations to increase motion legibility and convey a hand-made look. By analyzing the motion flows produced by these two approaches, we will discuss which structural motion properties may be responsible for the legibility, credibility, and specific look-and-feel of traditional 2D animations. In particular, we hope this will draw the attention of our fellow vision scientists toward the host of visual cues conveyed by the animation medium, and motivate new perceptual studies.