Three different softwares were developed to cover the stages involved in the process:

1. 1.A light render to generate synthetic images and record light position and camera parameters for each acquired image (not fundamental but allowed more control over the generation process)
   

2. 2.A light PTM fitter two convert all the different images from a single view point with different light direction into a single PTM file. Basically the fitter loads all the images from a single viewpoint and the corresponding light direction information. The fiiter uses a least square fit technique to fit the samples to the polynom, writing the result to a single file.
   

3. 3.A viewer that loads all the PTM and corresponding camera positions and transformation matrix (view and projection) and generates interactively a new image. The operation is done in two stages:
   

1. a.    View interpolation : the desired viewpoint rays are traced and intercepted with the planes (s,t) and (u,v), the closes existing ray is selected and the PTM coefficients are stored to a buffer.
   

2. b.    Light interpolation: the buffer produced in view interpolation is sent to the graphic card and a CUDA kernel evaluates the polynom at each pixel for the desired light direction.