Cluster of hundreds to thousands of CPUs/GPUs computing frames in parallel — distributes jobs across nodes, cuts wait time dramatically. Costs electricity and hardware per frame.
Render Farm
Hundreds or thousands of computers working in parallel on a job — that's the basic idea, and on set or at a VFX house, you immediately notice whether your infrastructure can handle it or not. A render farm isn't a single device, but a cluster of CPUs and GPUs coordinated via a network and job management software. You send off a frame or a sequence, and the farm breaks it down into tasks, distributes them to available nodes, and writes the finished material back to storage.
In practice: You're sitting with your 3D lead at the VFX house, and rendering a complex shot takes 12 hours on a single machine. With a 100-node farm, the same shot theoretically runs through in under 7 minutes — provided the software and network don't become bottlenecks. That's the crucial point: a poorly configured farm with slow hard drives or outdated job queue software is more of an obstacle than a solution. You need redundant storage systems, fast network connectivity (10Gbps minimum), and a robust render engine like RenderMan, V-Ray, or Houdini COPS that reliably distributes jobs. Each node calculates locally, caches textures and geometry, and stores the result centrally. Failures of individual machines have minimal impact on the overall time — the farm redistributes the load.
The economic factor: Every frame costs you electricity. For large productions (photo-realistic VFX, CGI-intensive sequences), render farms can mean millions in costs. Small studios therefore often use cloud render services or invest in GPU-accelerated rendering — NVIDIA RTX has become standard here because GPUs utilize throughput per watt significantly better than older generation CPU clusters. The trend is moving away from the pure CPU farm model towards hybrid setups: CPU nodes for complex simulations and lighting, GPU nodes for fast sampling and denoising.
Practically, this means for you on set: with a modern farm, you have revision passes in hours instead of days. You play in the final render in the evening, and in the morning, you have material for review. This changes the entire workflow — decisions can be made faster, and changes no longer cost the full production time. At the same time, you need someone to keep this infrastructure alive: a render wrangler, a sys admin, someone who debugs jobs and resubmits lost frames.