| 摘要 |
<p> Stereoscopic or multiple viewpoint images are generated by identifying individual objects within a 3-D virtual environment and rendering individual image segments for those objects in isolation (14). A first half of a stereoscopic pair is formed by simply compositing (18) the object image segments at their original locations before output to buffering (22) and display means. The second image of the stereoscopic pair is generated by applying positional, and optionally rotational, offsets to the same set of object image segments: the amount of offset being determined (20) by the object's apparent depth in relation to perspected depth and spacing of the first and second viewpoints. The second image is then composited from the variously offset object image segments. Further images, as for a multiple view autostereoscopic display, are obtained from further offsetting and compositing operations. An implementation feature, for generating defocussing effects from variation in viewpoint spacing and/or perspected depth modification, is also described.</p> |
