One of the most successful algorithms that bring realism to the world of 3D image generation is Phong shading. Phong shading can represent curved surface by smooth shading meshes of planar polygons. The level of realism that can be obtained by Phong shading is attractive for 3D applications. But, Gouraud shading has been used instead of Phong shading because of per pixel computation and hardware costs. However, with continuous improvement of VLSI technologies and request of higher realism, real-time Phong shading will be the next technology-push in 3D graphics.
Over the past few decades, many attempts have been made to make Phong shading practical for hardware implementation. One of typical methods was introduced : Taylor series approximation.
In this paper or dissertation, we determined hardware size of Taylor series approximation in terms of hardware minimization through visual comparison in a 3-stage pipelined architecture.
We minimized hardware size by reducing ROM table while keeping visual quality. ROM table size of uniform quantization is 256x4096x8bits. We minimize ROM table to 1/64 using adaptive-compressed non-uniform quantization with an exponent.
We also designed Taylor series approximation dedicated hardware successfully through minimization of hardware size by reducing ROM table while keeping visual quality.