MEDIA & ENTERTAINMENT
The render farm,
reinvented.
GPU workstations for artists, in a browser tab, on hardware you already own.
“If a user can connect to the web, they can reach their workstation on our secure network.”
DONALD STRUBLER · HEAD OF TECHNOLOGY, R3D STUDIOS · 2025
R3D STUDIOS · RESULTS FROM THEIR TIME ON ORION
60 sec
Workstation provisioning during R3D’s run
2:1
GPU density during R3D’s production run
Every seat, every frame, one cluster.
A supervisor can join any session.
Review sessions, direct links, and every protocol your artists already use, from one workloads table. Artists can connect from a tablet through NICE DCV.
PER-SHOW WORKSPACES
Spin up for the show. Hibernate at wrap.
Each show gets its own workspace, quotas, and mounts. Quotas keep a deadline from starving the rest of the studio.
Per show
Activate for the show, hibernate at wrap.
GPU slicing
Run more artist sessions per GPU. Time-slicing shares one card across several container workstations, so lighter work like comp and editorial doesn’t need a GPU per seat.
Burst
Add cloud nodes for the crunch, drop them after.
R3D STUDIOS · RESULTS FROM THEIR TIME ON ORION
60s
workstation provisioning during R3D’s run
Zero
critical failures during R3D’s production run
Up to ~40%
compute cost reduction during R3D’s run
Questions studios ask
More from the studio side.
Posts filed under VFX, newest first.
KEEP GOING
