Get started
Render your animation on a rented GPU
- 1
Get the project ready to travel
In Blender use File > External Data > Pack Resources (or keep every texture in the project folder with relative paths). Set the frame range, resolution, samples and output format in Output Properties, set Cycles as the engine, and save.
- 2
Rent a GPU machine
On RunPod, deploy a GPU pod with a plain Ubuntu or PyTorch template, add a volume disk for your files and choose Secure or Community cloud. An RTX 4090 was $0.34 (community) or $0.89 (secure) an hour on 7 Nov 2026. On another GPU cloud, pick any Linux machine with an NVIDIA card and SSH access.
- 3
Install Blender on the machine
Download the Linux build from blender.org/download onto the machine and unpack it with tar -xf. No installer and no screen are needed. If Blender stops on a missing system library, install it with apt (libxi6, libxrender1, libxkbcommon0, libsm6 and libgl1 are the usual ones).
- 4
Upload the project
Copy the .blend file and its folder to the volume disk (on RunPod, /workspace) with scp, rsync, runpodctl send and receive, or the Jupyter upload button. Keep it on the volume so it survives a restart.
- 5
Render one test frame
Run ./blender -b scene.blend -o //out/frame_#### -F PNG -f 1 -- --cycles-device OPTIX and open the result. Check textures, lights and the time it took before you pay for the whole range. Use CUDA instead of OPTIX on cards without RT cores.
- 6
Render the animation
Run ./blender -b scene.blend -o //out/frame_#### -F PNG -a -- --cycles-device OPTIX inside tmux or with nohup, so a dropped connection doesn't stop it. Blender reads arguments in order, so -o and -F go before -a. Add -s and -e to render part of the range.
- 7
Download and encode
Download the out folder, then join the frames: ffmpeg -framerate 24 -i out/frame_%04d.png -c:v libx264 -pix_fmt yuv420p -crf 18 shot.mp4. You can also encode on the machine first and download one file.
- 8
Stop billing
Stop the pod when the frames are safe on your computer, or terminate it to delete everything. On RunPod a stopped pod still bills its volume disk at $0.20 per GB per month; a terminated pod bills nothing but loses data not on a network volume.
The commands, in one place
Render the whole animation with Cycles on OptiX (from the Blender manual's command line options; the Cycles device goes after a double dash):
Headless render
./blender -b scene.blend -E CYCLES -o //out/frame_#### -F PNG -a -- --cycles-device OPTIX
Join the frames into a video
Run this where the frames are. Change -framerate to your scene's frame rate.
Encode with ffmpeg
ffmpeg -framerate 24 -i out/frame_%04d.png -c:v libx264 -pix_fmt yuv420p -crf 18 shot.mp4
Which GPU to rent for a first render
Prices checked 7 Nov 2026. They change often, so check the provider's own page before you spend. Stock changes by the hour, so a card may be unavailable when you look.
Prices checked 7 Nov 2026, on-demand, per hour.
| GPU | VRAM | Secure | Community | Open Data score |
|---|---|---|---|---|
| RTX A5000 | 24 GB | $0.27 | $0.16 | 4,390 |
| RTX 4090 | 24 GB | $0.89 | $0.34 | 11,061 |
| RTX 5090 | 32 GB | $1.19 | $0.69 | 14,966 |
| A40 | 48 GB | $0.59 | $0.35 | 4,077 |
Sources, as checked:
- RunPod GPU pricing (runpod.io): runpod.io/pricing
- RunPod docs: pod pricing and storage: docs.runpod.io/pods/pricing
- Blender Open Data benchmark medians (opendata.blender.org): opendata.blender.org/
Cycles vs Eevee on a headless machine
Cycles on CUDA or OptiX renders without any display, which is why it is the usual choice on rented machines. Eevee draws with the GPU's graphics side, so on a headless Linux machine it needs EGL. Many container images don't ship it, and Eevee then fails or falls back to software. If you need Eevee, test one frame first or pick an image with EGL support.
What keeps billing after the render
RunPod bills compute and storage by the second (checked 7 Nov 2026):
- A running pod: the GPU price per hour, plus $0.10 per GB per month for container and volume disk
- A stopped pod: no GPU charge, but the volume disk bills at $0.20 per GB per month
- A network volume: $0.07 per GB per month (under 1 TB), whether the pod runs or not
- A terminated pod: nothing, and every file not on a network volume is gone
If the shot is for an ad, previs or a social clip
Sometimes you don't need the final Blender render at all. If the shot is previs, an ad or a social clip, you can plan it in Stage, KreatorFlow's free 3D shot planner in the browser, and press Make it real to turn the planned shot into an AI video. Stage starts from 14 example scenes, imports your own models as .glb, .gltf or .vrm (up to 50 MB), and has no render queue on your side.
Planning in Stage is free. Turning a shot into AI video spends credits only after you confirm: in the Animate dialog, a 10-second Seedance 2.5 clip is 47 credits at 720p and 100 credits at 1080p, from KreatorFlow's own credit code. Make it real, in Simple mode, makes the whole video at 720p and shows its exact credits first. Making the video needs a paid plan.
An AI video is not a frame-exact Blender render. The model invents surfaces, cloth, hair and small motion, and two takes from the same plan differ. It suits previs, ads and social clips, not a shot that has to match your Blender scene pixel for pixel. Stage does not open .blend files, so export glTF from Blender first.
Questions other guides skip
Do I need a desktop or VNC on the cloud machine?
No. blender -b runs without a window. You only need SSH (or the provider's web terminal) and somewhere to put the files.
OptiX or CUDA?
OptiX on RTX cards (it uses the RT cores and is usually faster); CUDA on older NVIDIA cards. Both work headless. The device is set after a double dash: -- --cycles-device OPTIX.
What if the render stops halfway?
Because you render to numbered PNG frames, re-run with -s set to the first missing frame. In Output Properties, turn off Overwrite (and turn on Placeholders) so frames that already exist are skipped.
Frequently asked questions
Can I render Blender on RunPod?
Yes. Start a GPU pod, download the Linux build of Blender, upload your project and run blender -b with -a. An RTX 4090 pod was $0.34 an hour on community cloud on 7 Nov 2026.
Does Blender need a display to render on a server?
Cycles on CUDA or OptiX doesn't. Eevee on a headless Linux machine needs EGL.
How do I stop paying after the render?
Stop or terminate the machine. On RunPod a stopped pod still bills its volume disk ($0.20 per GB per month); terminating it bills nothing and deletes the data not on a network volume.
Does this work on other GPU clouds?
Yes. Any Linux machine with an NVIDIA GPU, SSH and a disk will do: the Blender and ffmpeg commands are the same.
