I doubt you will find too many people here who have worked on this, maybe you could try one of the Chromium mailing lists? It is an active project, and it does exactly what you want to do (hell maybe there is something on their todo list which you could attempt). You should probably try asking on comp.graphics.algorithms too.

From my laymans point of view bandwith considerations are exactly why depth compositing has been pretty unpopular since the days of PixelFlow. Screen space subdivision seems usually more efficient. Though even with just a single full framebuffer worth of data to pass to a central PC for each fps you will probably want gigabit ethernet.

Hi,
r there any fast compression & decompression techniques for 3D as well as 2D data that will be transmitted over the network. This way the transmission over head can be lowered. The compression techniques must be fast enough so that it does not itself become an over head & become useless.
thanx
bye

For 2D there is Huffyuv, for 3D I dont think you should be thinking about compression ... every computer should have it''s own copy of the scene, only getting updates (preferably by multicast) and rendering commands over the network. With that kind of setup I doubt you would need compression for the vertex data.