Normal Maps to improve facial definition.

One of the new technologies being used by video game manufactures is normal mapping, A normal map is applied to a low poly gone model, and create added depth by using the normal map to effect the shading of the low poly gon model. it differ from a normal texture because the "normal map" is effect by the inverimetal lighting. A normal map can make a 1,000 poly gon model look like a 15,000 poly gon model. And unlike a plane texture the illation isn't broken by change the camera angle, or the lighting angle.

To create a normal map one must first have a high poly gone model, such as a face consisting of 15,000 poly gons. And then a low poly gone face, say one consisting of 1,000 poly gons. The process of creating a normal map is to look at the 15,000 poly gone model while comparing it to the 1,000 model. As the computer compares the two models, it look at how light hit the 15,000 model, and figure out how a texture can be used to create the illusion of 15,000 polys on 1,000 face.

I think normal mapping could be applied to the player models with little additional over head, and would drastically improve the look of the players. Giving them added detail, and smooth the shape of the body.

It could be supported by adding the ability to chose a normal map for players, or by incorporating the ability to use high poly gone model in the appearance editor.

I have included an example image, Witch is from tom's hardare http://www20.graphics.tomshardware.com/graphic/20040504/ati-x800-07.html .

How do you envision this working on SL's deformable meshes? A standard high-res mesh, fed the av parameters, the normal map generated, then applied to the lower-poly mesh?

Depends how the models were originally created.

If they were splines and such, then it's simple to scale/deform then map.