The Client
Afroframe Studios is a 3D animation studio in Victoria Island, Lagos. Founded in 2021 by Chidi Oduya and Fatimah Sule, the studio produces animated content for children's television, branded explainer content, and increasingly, short-form animated segments for streaming platforms. By 2026, they had a team of seven: three animators, two riggers/technical artists, a compositor, and a studio manager. They were producing four to six minutes of finished 3D animation per month — a respectable output for a team of their size.
Their bottleneck was their render farm. Or more precisely, the thing they called a render farm, which was actually two gaming PCs and one workstation running Blender's Cycles renderer overnight and on weekends. These machines produced the rendered frames that became the finished content. They were producing those frames far too slowly to support the studio's growth ambitions.
The Challenge
3D animation rendering is one of the most computationally intensive workloads that exists in commercial creative work. A single frame of a finished 3D scene in Cycles — a path-traced renderer — can take anywhere from 2 minutes to 4 hours to render, depending on scene complexity, lighting, resolution, and render settings. Afroframe was delivering at 1080p with full global illumination and subsurface scattering — a cinematic quality standard that requires real compute to achieve in reasonable time.
Their existing "farm" — two RTX 3070 gaming PCs and one RTX 3080 workstation — could collectively render approximately 45 frames per hour on their most complex scenes. A one-minute animation at 24fps is 1,440 frames. Their most complex sequences were taking over 32 hours of farm time per minute of finished animation. A typical 90-second branded animation took 48 hours of render time. They were spending every weekend rendering what they'd animated during the week.
The problem compounded when client revision requests arrived. A client asking for a lighting change on a 90-second animation meant re-rendering affected scenes — potentially another 24–48 hours. Meeting tight revision turnaround times was becoming impossible.
The Consultation
We spent time with Afroframe's technical director — Chidi himself, who handles the pipeline — understanding their Blender workflow, their typical scene complexity, and their growth targets. They wanted to double output in 12 months: eight to twelve minutes of finished content per month. With their existing render capacity, that was mathematically impossible without working around the clock.
The solution was a dedicated GPU render farm: nodes purpose-built for Blender Cycles rendering, with high-VRAM NVIDIA GPUs (Cycles is CUDA-optimised), no unnecessary components (no display output, no gaming peripherals, minimal CPU — the GPUs do all the work), managed by a render farm software package (we recommended Flamenco, Blender's own open-source farm manager).
We also discussed power and cooling. Running multiple high-TDP GPUs in a Lagos office requires thinking carefully about the electrical load, heat output, and what happens when NEPA cuts power mid-render. We designed for all three.
The Build
4 Render Nodes — ₦3.8 million each:
- GPU: 2× NVIDIA RTX 4090 24GB per node (8 GPUs total across the farm) — Cycles renders to VRAM first; 24GB per GPU handles extremely complex scenes
- CPU: Intel Core i5-13400 — minimal, just enough to manage the render process; the GPUs are doing 95% of the work
- RAM: 64GB DDR5 — scene data is loaded to RAM before being dispatched to GPUs
- Storage: 512GB NVMe (OS + Blender) + 10GbE NIC for receiving render jobs from the studio NAS
- Case: 4U rackmount chassis — installed in a dedicated server cabinet with cable management
- Cooling: High-static-pressure server fans — render nodes run hot and loud; they're in a separate rack room away from the studio floor
Infrastructure:
- 40kVA online UPS for the entire farm — keeps rendering through power outages up to 45 minutes
- 10GbE switch and structured cabling between farm nodes and studio NAS
- Flamenco render farm software configured and tested
- Temperature monitoring and automatic shutdown triggers if the rack room exceeds 38°C
Total farm investment: ₦17.2 million. Existing workstations retained as artist stations.
The Result
The numbers are stark. Eight RTX 4090 GPUs collectively render approximately 380 frames per hour on Afroframe's complex scenes — versus 45 frames per hour before. A 90-second branded animation that previously occupied the farm for 48 hours now completes in under 6 hours. Same day turnaround on client revisions — not a marketing claim, but the actual operational reality of the new farm.
In the three months since installation, Afroframe has delivered three projects that would previously have been declined because of timeline constraints. Their studio manager reports that the farm is running at approximately 70% utilisation — they have headroom to grow without another hardware investment for the foreseeable future.
Chidi's estimate: "We doubled our capacity without hiring anyone. That's not a small thing for a studio our size."
Key Takeaway
For animation studios, render capacity is revenue capacity. A render farm that keeps up with your artists enables you to take on more projects, meet tighter deadlines, and offer revision turnaround times that command premium rates. A render farm that can't keep up becomes the ceiling on your business growth. The economics of a proper GPU render farm — particularly with current-generation RTX GPUs — have shifted dramatically in the last two years. The investment pays back in billable capacity faster than most studios expect.
Running an animation or VFX studio that needs more render capacity? Talk to our team about a render farm consultation. We design, build, and install.