摘要 |
A computer-implemented technique allows a user to perform difficult tasks such as extracting and manipulating 3D objects in 2D images of 3D scenes, and placing modeled 3D objects in 2D images of 3D scenes. The system displays a wireframe cage of the desired type of 3D object. The user drags control handles of the cage onto corresponding vertices of the pictured object. The system responds by determining an orientation and position of the modeled object in 3D space, and projecting at least some elements of the modeled object onto a projection plane. Portions of the 2D picture are associated with visible surface portions of the modeled object as projected onto the image plane. As the user manipulates the object, as for example by translating or rotating it, the portions of the picture remain associated with the surface portions of the object in its new position or orientation. The result is that the user can, in effect, select the 2D representation of the object, pull it out of the scene, and manipulate it as a 3D object. |