Abstract
3D shape can be inferred from a variety of different visual cues. Here, the way surface reflectance, depth gain, and scene complexity may influence our perception of depth on 2D images was tested. The perceived 3D shape from images of household objects was measured using a gauge figure: the slant and tilt of which was adjusted by the participant, at multiple sample points on the surface of the target object. These settings were then used to create the pictorial relief for the objects. First, the effect of three surface conditions on perceived shape was assessed: 1) full colour and texture; 2) a grey-scale representation, which preserved texture and reflectance cues; 3) a uniform grey reflectance, which provided only shape from shading cues to global shape and surface texture. Following this, the effect of scene complexity was explored, to assess whether perceived depth of individual objects can be affected by the number of objects observed. Finally, the gain was varied by squashing and stretching the depth of the target object, to assess the extent to which observers could perceive metric depth from pictorial cues. The results show that perceived depth was closely related to the ground truth shape, but was not affected by the surface manipulations, or scene complexity. These results show that observers were able to make use of shape from shading and surface texture cues to perceive 3D shape. However, pictorial relief did not vary with the depth gain applied to the objects. These results show that, for these viewing conditions, pictorial relief was limited to a bas relief ambiguity, and observers were not able to recover scaled metric depth.