3DNeo, I think I understand what you are getting at and I will try to explain the low poly / V4 issue as best as I can in my inebriated state.

Sometimes throwing more brute force processing power won't solve the problem as effectively as streamlining the entire process. Said another way, sometimes less is more. Enter subdivision surfaces.

V4 has a slightly lower poly count than V3 - 68,000 polys verses 72,000 polys respectively or something like that. That many vertices are required for a polygon renderer to create "smooth" silouetted (sp) images (though if you viewed an alpha channel zoomed close enough you would still see facets). Several of these characters will kill your viewport's response and interactivity.

By contrast, take a 3000 poly character and subdivide it. Now you have a smoothly rendered, low weight character that refreshes very quickly in the viewport. This is the industry standard for feature films, and this is the high-end resource of which you speak, with local details provided by sophisticated displacement mapping.

When you are rigging using weightmaps, it is much easier to skin a 3,000 poly character and "wrap deform" it to a subdivision surface than to weight map a 70,000 poly character. There is no debate on this point.

Speaking with the utmost respect and humility, I think your heart is in the right place but your method is not fully realized.