Hi,
I am currently evaluating MakeHuman for use in our tool chain. The intended use is to create real-time meshes and normal maps for use in an opengl/directx application. Generating the normal map should be straight forward given the high quality of the mesh MakeHuman currently produces. A cursory glance over the code suggests the meshes are stored as subdivision surfaces which depending on the detail of the control mesh may make it easy to produce a lower quality mesh than the one currently exported by the Collada Exporter. I am looking at 5k polygons max per character, which is significantly less than the 41k polygons that the collada exporter currently produces. Naturally there is no need for the genitals, bone structure and other internals (which were included in the 41k polygon figure), so how hard will it be to remove these on export?
Pointers from developers on getting a start on this, especially the reduction of LOD would speed things along greatly but I don't expect the job to be done for us.
Any code we build would mostly be in python and would definitely be contributed back to the project as the plan is to distribute the tool to our users. I'd also like to say that my management is very hot on the idea of using MakeHuman as am I, we are very impressed with the quality of both the tool and (very importantly) code base.
Thanks!