VRay Proxy question

[armilla]

A quick question:

Is there any benefit to converting an object that is maybe 100,000 polys into a VRay proxy, even if it is going to be unique in the scene, ie. it's not going to be duplicated at all in the scene? Does it offer any performance benefits?

I always thought it was only worth using proxies if they were going to be duplicated at least a few times, but the place I'm currently working at seem intent on doing one off proxies of say a car that will only appear once in the scene.

Cheers.

[super gnu]

unless you are short of ram, there isnt really a benefit.. they usually render a touch slower than native geometrydue to being read from the disk/network.

if im working on a huge scene (i.e. giant masterplan) i will model up each building and then convert to proxy. this keeps the scene light to navigate and keeps the ram use under control.. but if youre not talking about vast scenes or multiple copies theres not a benefit really.

[armilla]

Thanks for the fast reply super gnu, that's pretty much as I thought. I can see the viewport performance benefit with a proxy, but you could also just keep the native geometry and "Display as Box", which would help viewport navigation and wouldn't slow things down at render time when the proxy is read from the network.

Crazy thing is, they're proxying massive one off objects and yet not proxying 20k poly objects that are in the scene multiple times. Bit of a bonkers workflow to my mind.

[super gnu]

there is actually a flipside to that argument.. if youre talking aboutlots of copies of an object thats a few thousand polys (i.e. an arch vis person model) its also generally better to either use straight max instancing, or proxy a big chunk of them in a single proxy. people often forget that instancing works very well with max/vray

i generally reserve proxies for massive scenes where i proxy almost everything apart from the landscape, and scenes where medium to high poly objects are repeated many times ( carparks, forests etc.. )