Hi,

I'm working on a 3D Software where I should be able to display a scene/model with several millions of elements, some models are 100 GB on hard drive.

So, I would like to know how softwares like 3DS/Maya/... support this kind of models ?
Do they play with a kind of LOD model, out of core system, simplified models etc... ?

Do you know all the way to do ?

If you know algorithms etc... it is great too :-)

Thanks

The most common way to conserve memory in such cases is to use proxy models when animating. I don't know if any common modeling software supports single model weighing at 100GB, though.

If you can split the models to fit available memory, you can render the pieces separately and composite the results together as layers. This is very common in modern movie rendering, as well as in modern games. In order to fit the data in GPU memory, you will have to split it anyway - the top-end GPUs of today have about 6GB of RAM total per board, including the frame buffers. Some hardware solutions let you stream geometry directly from system memory to GPU's input assembler, though.

100GB worth of geometry is hard to push over a graphics bus in realtime. Even if you have enough system memory to actually hold the mesh (which is unlikely but possible), you will meet various bottlenecks along the way. On commodity hardware, expect tens of seconds of render time per frame, especially if you have to use HDD for paging the data.

there was this paper that could be interesting

"Mega Meshes - Modelling, rendering and lighting a world made of 100 billion polygons".

http://miciwan.com/GDC2011/GDC2011_Mega_Meshes.pdf