Short answer: what animation rendering really costs is the render time per frame multiplied by the number of frames and by the hourly cost of the har
Short answer: what animation rendering really costs is the render time per frame multiplied by the number of frames and by the hourly cost of the hardware that does the work, plus the costs that do not appear on a render invoice: idle time, re-renders and your own waiting time. Rendering locally is never free, and the cheapest cloud option depends on how each service bills, not on the headline rate.
This guide shows what animation rendering really costs, with the formulas, a worked example with clearly marked assumptions, a comparison of how several render services state that they bill, and the points where animators most often overspend. Prices change often. Every figure taken from a service was read from its public pricing page on 30 October 2026 and should be rechecked before you budget a job.
What animation rendering really costs: the formula
To find what animation rendering really costs for your own project, start from one measured frame. Render a representative frame at final settings, note how long it takes, then scale up:
- Total render hours = (minutes per frame × number of frames) ÷ 60, divided by the number of machines rendering in parallel if you want wall-clock time.
- Cloud cost = machine-hours × hourly rate of the machine type, plus any billed idle time.
- Local cost = kWh used × your electricity price, plus hardware depreciation for the hours used.
- Revision factor = multiply the total by the number of full or partial re-renders you expect. A client round of changes can double the bill.
Test frames matter more than any estimate of what animation rendering really costs. Motion blur, volumetrics and heavy textures can make frames in the same shot differ widely in render time, so sample the easiest and the hardest frames, not only the first one.
Worked example with stated assumptions
The numbers below are illustrative inputs, not measurements. Replace them with your own.
| Item | Assumption | Result |
|---|---|---|
| Local render | One workstation drawing 800 W at the wall (the RTX 5090 alone is rated at 575 W total graphics power by NVIDIA), 72 hours, an example electricity price of $0.20 per kWh | 0.8 kW × 72 h = 57.6 kWh, 57.6 × $0.20 = $11.52 of electricity |
| Cloud render | 8 hours on a 4× RTX 5090 server at iRender’s listed 3+ hour rate of $34.20 per hour, billed per minute | 480 min = 8 h × $34.20 = $273.60 |
| Forgotten server | The same server left on for 45 minutes after the job ends | 0.75 h × $34.20 = $25.65 of pure idle cost |
The local figure looks tiny because electricity is only one part of what animation rendering really costs. Add the depreciation of the card (purchase price minus expected resale value, divided by the hours you expect to use it), the time you cannot bill while the machine is busy, and the cost of replacing hardware that fails. The two examples also do not render the same scene in the same time, so they show how to structure a comparison, not which option is cheaper. Use your own test frame to compare.
How render services say they bill
Billing units differ, so what animation rendering really costs depends on how each service counts time, and a lower number per hour does not always mean a lower invoice. The table lists what each service states on its own pricing page. We have not measured render times on these services.
| Service | Billing basis as stated | Notes |
|---|---|---|
| SheepIt Render Farm | Free, community powered. Described as for Blender (Cycles, Eevee, Workbench) | Funded by donations. Jobs run on volunteer machines, so turnaround is not guaranteed |
| Fox Renderfarm | CPU rendering billed per node per hour; listed tiers from $1.224 (standard) down to $0.734 (highest recharge tier) | The pricing page showed no GPU rates when we checked. A $25 trial credit is advertised |
| GarageFarm.NET | Billed by the second of active rendering; CPU priced per GHz·hour, GPU per OB·hour, with Low, Medium and High priority tiers | Queue waits and software installation are stated as not billed. Prices exclude VAT |
| iRender | Dedicated GPU servers billed per minute of active use; sessions of 3 hours or more get about 10% off the list rate. Examples: 2× RTX 4090 at $15 per hour, 4× RTX 5090 at $38 per hour list | Server-based (IaaS) model, so you pay while the server is on, whether or not it is rendering |
| RebusFarm | Not included | Its pricing page could not be retrieved when we checked, so we state no figures |
Service models matter here. Fox and GarageFarm are managed farms where you submit a job and the service schedules it, so you pay for rendering time. iRender provides remote servers that you operate yourself, so you pay for the time the server is on and control how well that time is used. The guide to IaaS, SaaS and PaaS render farms explains the difference in more detail. iRender states that it holds ISO 27001 (BS7799), ISO 20000 (ITIL) and ISO 9001:2000 certification, Uptime Tier III data centres, and applies a non-disclosure agreement as set out in Section 14 of its terms and conditions. This site is editorially independent and does not verify these claims.
Local rendering is not free
No invoice arrives when you render at home, which makes the cost easy to ignore. It still includes electricity under sustained load, depreciation of hardware that loses value whether you use it or not, and the hours when your main machine is tied up. The article on the hidden costs of rendering locally goes through each of these, which is the part of what animation rendering really costs that people forget. On a long project, the local cost depends on how much of the time the machine is actually rendering. Local rendering usually wins when you render constantly and your hardware fits your scenes. Cloud rendering usually wins for short bursts, deadlines and scenes that exceed your own GPU memory.
Where what animation rendering really costs gets out of hand
| Cost | Where it hides | How to reduce it |
|---|---|---|
| Per-frame render time | Excess samples, oversized textures, unnecessary motion blur or volumetrics | Optimize on a test frame before submitting the full range |
| Idle server time (hourly billing) | The server stays on after the render, or while you upload, set up and download | Prepare assets first, shut down immediately, use auto-shutdown if available |
| Failed or wrong frames | A wrong setting found after hundreds of frames have rendered | Render a short test range, then check frames across the shot before the full job |
| Re-renders | Client revisions that change the whole sequence | Render in layers or passes where the pipeline allows it, so a change needs fewer frames |
| Transfer and storage | Large caches and texture sets upload and download slowly | Pack only what the scene uses, and check storage and transfer terms before you commit |
For tactics in depth, see cutting a render farm bill and why the first farm bill is higher than expected. To turn the formula above into a number for your own project, use the render farm cost calculator.
How to choose without guessing what animation rendering really costs
- Render one worst-case test frame and write down the time and the memory it needs.
- Check that the service supports your software and renderer versions and any plugins.
- Convert each service’s billing unit into the cost of that one frame, then scale to the full shot.
- Add idle time, re-renders and transfer time before comparing totals.
- Run a short paid test on your first choice before you commit a deadline to it.
Our comparison of render farms and the guide to saving money on a render farm help with the shortlist.
FAQ
How do I estimate the cost of rendering an animation?
Render a representative frame at final settings, multiply its time by the frame count, then multiply by the hourly cost of the machine. Add idle time, a revision allowance and any transfer or storage charges. Repeat with an easy and a hard frame to get a range.
Is cloud rendering cheaper than rendering locally?
It depends on utilisation and on what you count. Constant rendering on well-matched hardware usually favours local machines. Occasional heavy jobs, tight deadlines or scenes larger than your GPU memory usually favour the cloud. Compare using your own test frame.
What is the difference between per-frame, per-hour and per-second billing?
Managed farms often bill for the compute time a job uses, in units such as node-hours or GHz-hours. Server rental bills for the time the machine is on, here per minute. Billing by active rendering time does not charge you for idle gaps, while server rental does, so idle time becomes your responsibility.
Do render farms charge for idle time?
It depends on the service model. GarageFarm states that it bills by the second for active rendering only. iRender states that it bills per minute of active server use, and a running server counts as active. Check the terms of any service before you start a job.
Sources
- NVIDIA, GeForce RTX 5090 specifications (575 W total graphics power)
- iRender, pricing page, read 30 October 2026
- Fox Renderfarm, pricing page, read 30 October 2026
- GarageFarm.NET, pricing page, read 30 October 2026
- SheepIt Render Farm, home page, read 30 October 2026
- U.S. Energy Information Administration, Electric Power Monthly, for current average electricity prices in your region
COMMENTS