We present a method for sharing visual information in 3D virtual environments, using a projective texture mapping based method. Avatars can share information with other avatars by projecting relevant information into the environment. In addition to standard projective texture mapping, an important depth cue is added: projected light is attenuated in function of the light-travel distance. This is efficiently accomplished on a per vertex basis by adaptively super-sampling under-sampled polygons. This way, the projection quality is maximized while keeping a fixed frame rate. Our technique is implemented into the Quake III engine, extending its shading language with GLSL fragment and vertex shaders. Categories and Subject Descriptors: H.5.1 [INFORMATION INTERFACES AND PRESENTATION (I.7)]: Multimedia Information Systems, I.3.7 [COMPUTER GRAPHICS]: Three-Dimensional Graphics and Realism General Terms: Algorithms, Performance.