A lighting model processing system for a computer graphics workstation's shading function includes multiple floating point processing stages arranged and operated in pipeline. Each stage is constructed from one or more identical floating point processors. The lighting model processing system supports one or more light sources illuminating an object to be displayed, with parallel or perspective projection. Dynamic partitioning can be used to balance the computational workload among various of the processors in order to avoid a bottleneck in the pipeline. The high throughput of the pipeline system makes possible the rapid calculation and display of high quality shaded images.