Any form of adaptive tessellation?

I’m rendering a large, animated, and detailed ocean. I’ve already used as many tricks as I can find to balance rendertimes with detail, and I’m frankly not impressed.

Games like AC:Rogue and pretty much any Cryengine game have fantastic water and oceans that run in real-time. The only way this water can look so nice is adaptive tessellation. That is to blender users, batch or per-face subdivision based on distance to camera.

Blender can handle pretty much every visual element of the ocean gracefully except the geometry.

I’m aware that Blender is slowly moving towards OpenSubDiv, and eventually OpenGL2.1, which will help immensely, but even then, we will still need and benefit from adaptive subdivision.

I have seen both the old tessellation modifier and kostack’s modifier, both throw aside, and kostack’s doing exactly what I want. I can’t actually use it though as it’s for Blender version 2.49.