Proxy in blocks + proxy vs. instance (VfR Next)

[Micha]

Hi,

I have a train interior with a lot of seats. The seats based at one embedded block. I edit the embedded block and convert the geometry to a proxy. Than I press ok and all instances got the proxy preview model, but at the rendering no seats are visible. It looks like proxies in blocks doesn't work. I hope it's a bug and can be fixed. Blocks are very important, since it allow to easy change the blocked geometry.

General I ask me, is there an advantage to use proxies instead blocks for the render process? Doe's the GPU rendering profit by proxies more than using block/instances? Or are proxies good for keeping the scene light at Rhino only, but during the render process there is no advantage?

-Micha

[georgi.georgiev]

Hi,

In respect to the BlockEdit issue, this has been logged and our developers will further investigate if a solution could possibly be implemented.

Regarding advantages of V-Ray Proxies and Blocks, an improvement will be introduced in the next official update (already part of the latest stable night) and true instancing will work correctly. This means that a block geometry would be loaded once rather than the number of times it has been instanced. You will notice that memory usage will be optimized due to this change.

Specifically for V-Ray Proxies (vrmesh & vrscene), the main advantage is memory allocation. For more regarding this topic, I think you will find this thread useful:
https://forums.chaosgroup.com/forum/...usage-in-v-ray

The choice of using either V-Ray Proxies or Rhino Blocks comes down to a trade-of between viewport performance and the ability to keep the instanced objects fully editable.

[Micha]

Great to hear about the block/instancing improvement. Is the improvement good for CPU and GPU rendering, for RAM and VRAM?

I afraid I don't understand the details from the linked page. There is a lot written about RAM usage. What do you think about GPU rendering and GPU memory?

My scene works quite good at the Rhino viewport and I can use a 1080ti card for the viewport only, rendering could be done at two 2080ti. Can I save GPU memory if I use proxies instead instances? Or is there no difference if the scene is loaded to the GPU?