Multi-view Rendering using GPU for 3-D Displays

Creating computer graphics based content forstereoscopic and auto-stereoscopic displays require renderinga scene several times from slightly different viewpoints. In thatcase, maintaining real-time rendering can be a difficult goal if thegeometry reaches thousands of triangles. However, similaritiesexist among the vertices belonging to the different views likethe texture, some transformations or parts of the lightning. Inthis paper, we present a single pass algorithm using the GPUthat speeds-up the rendering of stereoscopic and multi-viewimages. The geometry is duplicated and transformed for thenew viewpoints using a shader program, which avoid redundantoperations on vertices