by paride4331 » Tue Jul 25, 2017 8:14 am
paride4331
Tue Jul 25, 2017 8:14 am
Hi coilbook,
Each moveable proxy object is "sent" to GPU individually; non-movable proxy objects are "sent" to GPUs in a "single geometry block".
To "build" this "single geometry block" Octane takes longer time to send movable proxy objects, but when it is into GPU,
Octane will be fast rendering frames after frames without voxels geometry each time.
In a scene with movable proxy objects and non-movable proxies:
All moving or modified objects (movable proxy) will be send to GPU each new frame,
but not non-movable proxy objects, already stored in VRAM.
Example:
10 movable proxy objects + 100 non-movable proxy objects:
Frame 1> first voxelization = 10 single objects to GPU + 100 objects built in "single geometry block" + 1 "single geometry block" to GPU
Frame 2> second voxelization = 10 single objects to GPU
Frame 3> third voxelization = 10 single objects to GPU
Regards
Paride
2 x Evga Titan X Hybrid / 3 x Evga RTX 2070 super Hybrid