So, real sappy but here's something that kept me up at night Not last night, had a pretty busy week so I kept it shelved until I had some time (now)
Anyway; in simplest terms the idea has been to improve the concept of the Normal Map by removing its biggest restriction: reliance on the face it's applied to being hit by the lighting. (one could argue the failure to cast shadows upon itself it a greater flaw but that has already been tackled)
The idea, to generate a Normal-Map-esque texture layer (can't use existing normal maps because it needs twice the colour range) which instead of taking the light collected by the surface and working from there - collects its lighting information from a single central point. This point is fixed in place and orientation to its model.
The light intensity, direction and colour are measured by this central point, and projected on a ''sphere map'' (I know this is an existing term - I do not seek to reference this definition, but couldn't (yet) think of a fitting term that isn't already in use) which indicates where on the texture the light is being received.
So what are the advantages over standard normal maps? Nowadays, not much. But it excels (theoretically) in providing excellent, dynamic lighting onto very-low-poly or even Two-Dimensional objects. It should be capable of phong-like highlights too, and even specular reflections. Though I haven't given that as much thought yet.
And the drawbacks? Well, there's plenty. It won't cast shadows, receive shadows, etc. Though it's not like this is intended for use in anything where that would have been a factor regardless. Oh and can't be used on any ''soft body'' models, or really anything where the object's polygons might at any point alter the relative position or orientation. Unless you assign the object multiple light points, obviously.
Hey, anyone got a time machine so I can think of this 20 years ago? Heh. Here's to hoping it doesn't already exist hm? Wouldn't that be embarrassing.