Image-based non-photorealistic rendering for realtime virtual sculpting

This paper discusses a real-time virtual sculpting system in a non-photorealistic manner. Up till now, a majority of researchers pursuing virtual sculpting have highlighted the physical fidelity of simulation, with approaches for realizing photorealism being principal specialized contributions. We aim at capitalizing on the strengths of non-photorealistic rendering with respect to aesthetics, expression and computing expense. By means of touch-enabled manipulation, the haptic interaction is conducted to form the deformable surfaces, and the edge extraction is employed to stress boundaries. Afterwards, we utilize anisotropic diffusion to lessen unimportant details as well as a fresh inverse Gaussian bilateral filter for spot removal. Processed in a spatiotemporal style, the non-photorealistic rendering works to maintain coherence in time. Moreover, the parallel realization of the system put forward on graphics hardware (GPU) guarantees real-time behavior.