It only takes normals and standard world & world*view*projection matrices in the vertex shader, and passes world position & normal in to the pixel shader, then the pixel shader takes light position, radus, intensity & color and computes the intensity by simple n.L formula (it takes into account light's radius and distance from the pixel). Now, the one part of the test sphere is always black, even when the light is really close to it and actually the whole lighting seems to be a little weird. I cannot find the reason for this, I tried to multiply normal with inverted world matrix (instead of world matrix) and to use vertex instead of pixel lighting, or even inverting the light vector, discarding last row and column from world matrix, but no luck.Here is the example with the code: http://www.sendspace.com/file/tjxldf
(in the code, include and lib files' path is specified absolutely because of mess on my hard drive)
PS: in the example .exe file, you will be in the sphere, move a little back or ahead to see it from outside, and the black line is light's direction (line start is the light position, and the end is position*10, i was to lazy to attach a billboard to it
Any clues, this is driving me crazy. But seriously.