Shape-from-shading is a wellknown computer vision problem which aims at computing the shape of a scene using one greylevel image of this scene only. In the simplest case, its model is the eikonal equation. When several images of the same, scene lighted from different directions, are available, the problem is called photometric stereo. In my talk, I will quickly survey the main pecularities of shape-from-shading and photometric stereo, and show some recent results obtained using both these techniques.