| 摘要 |
A method, system, and computer-readable storage medium are disclosed for generating a three-dimensional (3D) shape based on a two-dimensional (2D) image. In one embodiment, an image comprising a 2D representation of a 3D object is received. Characteristics of one or more light sources affecting the image may be determined. A 3D mesh approximating a shape of the 3D object may be generated based on a first downsampled version of the image. A candidate image may be rendered based on the 3D mesh and the characteristics of the light sources. The 3D mesh may be output if a difference between the image and the candidate image is smaller than a threshold error value.
|
