Cloud rendering solutions, built around how you actually work
Every project has a different bottleneck. These are the three shapes of work the farm is tuned for — pick the one that matches yours, or jump straight to your software.
Architectural Visualization
Photorealistic interiors and exteriors, where a single frame can hold millions of polygons of foliage and furniture and every surface needs true light transport. Stills go out on a handful of fast cards; a walkthrough goes out wide across many.
- Optimized VRAM allocation for large polycounts
- Flicker-free animations with pre-baked global illumination
- Works with SketchUp, Revit, Rhino and ArchiCAD models
VFX & Cinematic Animation
Sequence work, where the job is hundreds or thousands of frames and the bottleneck is throughput rather than any single frame. Rent more nodes and the sequence finishes sooner; the scene does not change.
- Full support for multi-pass OpenEXR output
- Configured for particle physics and volumetric simulation
- Compatible with Redshift, Arnold, Corona and V-Ray shaders
Studio API & Automation
Submission from your own tools rather than from ours — renders triggered on approval, projects synced from studio storage, frames pulled back into an asset manager without anyone opening an application.
- Secure SFTP file ingestion
- Webhooks for job start, completion and failure
- Dedicated servers for sensitive production work
Choosing hardware is part of the work
Most render farms decide this for you: you submit, you wait in a queue, and whatever machines are free are the machines you get. Here the hardware is a decision you make per job, and it is usually the decision that determines both the bill and the deadline.
A few frames, quickly
A hero still or a client revision wants the fastest card you can get and only one or two of them. Rent an RTX 5090, render, stop. Paying for a wide spread of nodes buys nothing when there are six frames to place on them.
A sequence, overnight
Eight hundred frames with no deadline until morning is the case where older cards win. An RTX 3090 costs a fraction of a 5090 per hour, and twenty of them running while you sleep finish long before you wake up.
A sequence, before a deadline
When the delivery is in two hours, width is what you are buying. Take the fast cards and take many of them — the same scene, the same settings, a different number on the node count. It costs about what two nodes would have: the frames are the same frames, and nodes are metered by the second, so you are buying the finish time and not the work twice.
This is the Blender and Unreal Engine service. Every node shows its GPU, VRAM, CPU, disk, network speed and hourly price before you rent it — a rate for comparing cards, billed by the second you actually use — and a render job can rent for you automatically — cheapest first, up to your balance — if you would rather not choose. Either way there is no queue to sit in: the machines are rented when you submit, so the only wait is the second it takes to rent them. Maya, 3ds Max and Cinema 4D work differently: the farm assigns the machines and bills the CPU and GPU hours the job uses, with no hardware for you to pick. Estimate a scene or read how the pricing works .
Rendering for your software
Dedicated setup guides, supported render engines and pricing for each application.
Need something that is not listed? Talk to the team about a dedicated pipeline.