For example I have a mesh whoes vertexs are all in a real far position from the origin, like (0, 0, 30000). When I render this mesh, I'll have a decimal number data precision issue in subdivision and subsurface scattering.
In game engines it's really easy to customize that transform all the renderable objects into camera space by multiply a view matrix, but I don't know how to achieve this in arnold.
My current solution is transform all my objects to camera space in DCC software. But this method cost a lot of extra work time and extra memory to store. As for arnold procedure nodes like some hair plugins' cache, I have to export them again using camera space mesh cache.
I know it's not good to create objects so far, but I have already done a lot of work which is unreversable. So I wonder how I can transform objects more efficiently. Please help.