Light is a kind of medium that makes objects visible without being visible itself in empty space. The combination of spatial and spectral characteristics of the light source, material reflectance and shape of an object determine its appearance, and as such the appearance of an object is the main cue for perception of its lighting. In the current study we tested whether types of shapes and scene layout influence lighting perception. In real scenes composed of several objects we tested observers' perception of lighting direction with a real light probe – using a novel experimental setup to optically mix the probe object into a real scene by a semi-transparent mirror. Participants (N=15) had to adjust the illumination direction on the probe such that it fitted the scene. We tested nine lighting directions for four different scenes, existing of five objects: a cylinder, star shape, and cross shape plus combinations of bowling pin(s) and pentagon shape(s). When using one bowling pin and one pentagon shape, the result showed that the light direction estimation was systematically contracted near the pentagon body, but not near the bowling pin. When horizontally mirroring this scene, also the light direction estimation was mirrored. Replacing the bowling pin with another pentagon body resulted in systematic contractions of the light direction slants near both pentagon bodies. Replacing the pentagon bodies with two bowling pins finally resulted in close to veridical light direction settings. The globally convex smoothly curved bowling pins, in comparison with the facetted pentagon shapes, improved observers' setting of light direction. Mirror arrangements of objects in the scene improved the estimation of the tilt direction. These very systematic effects suggest that human perception of the global "light flow" in a scene might be systematically deformed depending on scene layout and content.@en