Blender Not Rendering? 8 Causes and How to Fix Them (2026)
When Blender will not render, the cause is almost always one of a few settings, checked in this order:
- Camera and device: no active camera stops the render; a GPU backend must be chosen in Preferences and GPU Compute set in each scene.
- GPU memory: the most common reason a scene renders on CPU but not on GPU is that it does not fit in VRAM (Blender Manual).
- Output: F12 still renders are not saved automatically; animation renders are written to the output path (Blender Manual).
- On iRender: 1x to 8x RTX 4090 (24 GB) or RTX 5090 (32 GB) servers where you open the .blend and test one frame before the full range.
You press F12 or Ctrl+F12 and nothing useful happens: an error, a render that never starts, a GPU that is ignored, or a render that finishes but leaves no file. Blender rarely fails for no reason. Each of these symptoms points to one stage of the render.
This guide follows a render from Preferences to the saved file, with a table to match your symptom, a short script that prints the settings involved and a command-line render for when the interface gets in the way.
Checked against the Blender Manual (4.x to 5.3), November 2026.
1. Match your symptom
Start from what Blender shows you. Each row points to one setting:
Blender not rendering: symptom, cause, fix
Match what you see to a row, then make the change in the last column.
| What you see | Likely cause | Fix |
|---|---|---|
| “Error: No camera found in scene” | The scene has no active camera | Scene Properties › Camera, or select a camera and press Ctrl+Numpad 0 |
| GPU Compute is greyed out, or only CPU is listed | No GPU backend chosen in Preferences | Preferences › System › Cycles Render Devices: choose CUDA or OptiX and tick the GPU |
| Backend is set but the render still uses the CPU | The scene Device is still CPU | Render Properties › Device: GPU Compute (set per scene) |
| Renders on CPU but fails or stops on GPU | Scene does not fit in GPU memory | Smaller textures, simplify geometry, close other GPU apps, or more VRAM |
| Stuck on “Loading render kernels” | Kernels being loaded or compiled, or a driver problem | See our guide on Loading render kernels (linked below) |
| F12 finishes but no image on disk | Still renders are not saved automatically | Image › Save As in the render window, or render the animation |
| Ctrl+F12 ends at once or skips frames | Files already exist and Overwrite is off, or placeholders from another run | Output Properties: check the path, Overwrite and Placeholders, and the frame range |
| Interface freezes while rendering | The same GPU drives the display and the render | Render from the command line, or use a GPU that is not driving the monitor |
If your case is not in the table, open the system console (Window › Toggle System Console on Windows) before you render. Cycles prints the stage it is in and the error it stopped on.
2. Where a render stops
A render goes through the same stages every time. The first stage that fails decides what you see, so check them from left to right:
How to read the figure. The six boxes are the stages every Blender render passes, in order: render devices in Preferences, the scene’s device, the active camera, GPU memory, kernels and the output file. Under each stage is the symptom you see when it fails, such as the GPU missing from the device list, “No camera found”, Out of memory, a render stuck on “Loading…” or no file on disk. Match your symptom to a red box, fix that stage, and remember that a fix further right does nothing while an earlier stage still fails.
- Preferences: GPU rendering is enabled in Preferences › System › Cycles Render Devices, by choosing CUDA, OptiX, HIP, oneAPI or Metal and enabling the devices (Blender Manual).
- Scene: each scene must also be set to GPU Compute. A new or appended scene can still be on CPU.
- Camera: without an active camera there is nothing to render.
- Memory and kernels: covered in sections 3 and 4.
- Output: covered in section 5.
3. The GPU is not used or not listed
If your card does not appear under Cycles Render Devices, check that it meets the minimum for the backend. For NVIDIA cards, the Blender Manual lists:
- CUDA: Windows or Linux, compute capability 5.0 or higher.
- OptiX: Windows or Linux, compute capability 5.0 or higher and a recent driver. The minimum driver changes with the Blender release; the Blender 5.2 LTS manual lists version 575 or newer.
A card that meets these but still does not appear usually needs a clean driver install. Check with the script below which backend and devices Blender actually sees:
If the list shows your GPU as off, tick it in Preferences and save the preferences. If Device prints CPU, switch the scene to GPU Compute.
4. It renders on CPU but not on GPU
The Blender Manual gives the most common cause as not enough memory on the graphics card. A GPU starts with only its own VRAM, which is usually much smaller than system RAM:
How to read the figure. Each bar is the scene’s memory measured against one GPU’s VRAM. In the top bar the scene fits and the GPU renders at full speed. In the middle bar it is slightly over: CUDA, OptiX, HIP and Metal move the rest into system RAM (the amber part), which is slower but usually still faster than the CPU. In the bottom bar it is far over, system memory cannot cover it either, and Cycles stops with “Out of memory”. The fixes under the bars apply to the middle and bottom cases.
With CUDA, OptiX, HIP and Metal, Blender tries system memory when VRAM is full. That is slower but still renders. When even that is not enough, Cycles stops with Out of memory. Reducing texture resolution is the first fix the manual names. See our guide on Blender Cycles out of GPU memory for more.
If the render hangs before the first sample on a new Blender version or a new driver, read how to fix Blender stuck on loading render kernels.
5. It renders but nothing is saved
This one looks like a failure but is how Blender works. The Output panel decides where renders go:
- Still renders (F12) are not saved automatically. Save them from the render window with Image › Save As, or write them with a File Output node.
- Animation renders (Ctrl+F12) are written to the output path. A path starting with // is relative to the .blend file.
- Overwrite off means frames that already exist are skipped. A re-render can end at once because every frame is already on disk.
- Placeholders creates empty frames while rendering, so several machines can share one folder. Empty files left by a stopped run are also skipped when Overwrite is off.
If the interface freezes or the display driver resets during long renders, render without the interface:
A command-line render uses the same settings saved in the .blend file, so save the file before you start it.
Blender render checklist
Tick each item before you render the full range.
SaaS or IaaS render farm when Blender will not render?
A scene that will not render at home because of a setting will not render on a farm either. A farm helps when the problem is the machine: not enough VRAM, an old GPU, or a computer you need back. On a SaaS farm you upload the scene and the farm renders it on its own setup. On an IaaS farm you rent the whole machine, open the scene yourself and fix it there.
Blender render problems on a farm
How the two farm models fit a scene that will not render.
| Your situation | SaaS render farm | IaaS render farm (iRender) |
|---|---|---|
| The scene works at home and you only need frames faster | A good fit: upload, and the farm splits the range across its nodes | Works too; you split the range across servers yourself |
| The scene runs out of GPU memory at home | Depends on the GPUs the farm uses | Pick a server with RTX 5090 (32 GB) or RTX 4090 (24 GB) and test one frame |
| You need a specific Blender version or add-ons | Depends on what the farm supports | You install them on the server once; they stay in your image |
| You have a deadline and need to know when rendering starts and ends | Jobs join the farm's queue, so the start time depends on how busy the farm is; many farms sell higher-priority plans to start sooner | The server runs only your job from the moment it boots, so you control when rendering starts and can time the whole job |
Rule of thumb: if the scene already renders correctly, a SaaS farm is the shorter path; if the problem is memory or setup, rent the machine and fix it there.
Test and render on an iRender server
How to read the figure. The left box is your own workstation and the right box is an iRender server; the rows between them stay the same: desktop, the Blender version you install, add-ons, licenses and render settings. Only the GPU count changes, to 1, 2, 4 or 8 RTX 5090 or RTX 4090 cards. The six-stage check above applies on the server too, starting with ticking every GPU under Preferences › System › Render Devices.
On iRender you rent a whole Windows server and connect by remote desktop:
- Upload the project through iRender Drive before you start the server; uploading does not use rental time.
- Start a 1x, 2x, 4x or 8x RTX 4090 (24 GB) or RTX 5090 (32 GB) server and install your Blender version once.
- Open the .blend, set the GPU backend in Preferences and run the script above.
- Render one frame and check the saved file.
- Render the range, from the interface or the command line. Multi-GPU servers bill per minute; 1-GPU servers have a one-hour minimum.
Frequently asked questions
1. Why is Blender not rendering anything?
Check the stages in order: an active camera, a GPU backend in Preferences, GPU Compute in the scene, enough GPU memory, and the output settings. The system console shows where the render stopped.
2. Why does Blender say “No camera found in scene”?
The scene has no active camera. Set one in Scene Properties › Camera, or select a camera and press Ctrl+Numpad 0.
3. Why is GPU Compute greyed out in Blender?
No GPU backend is enabled. Go to Preferences › System › Cycles Render Devices, choose CUDA or OptiX for NVIDIA, tick the GPU and save preferences.
4. Why does my scene render on the CPU but not on the GPU?
The Blender Manual names insufficient GPU memory as the most common cause. Blender can spill into system memory on CUDA, OptiX, HIP and Metal, but a scene far larger than VRAM stops with Out of memory.
5. Why did Blender render but not save my image?
Still renders (F12) are not saved automatically. Use Image › Save As. Animation renders (Ctrl+F12) are written to the output path in Output Properties.
6. Why does my animation render finish instantly?
The frames already exist and Overwrite is off, so Blender skips them. Turn Overwrite on, delete old frames, or change the output path.
7. Which NVIDIA GPUs work with Blender Cycles?
CUDA and OptiX need compute capability 5.0 or higher on Windows or Linux. OptiX also needs a recent driver; the minimum version is listed in the manual for your Blender release.
8. Can I render a Blender scene that is too big for my GPU?
Reduce texture sizes and simplify geometry first. If it still does not fit, render on a GPU with more VRAM, for example an RTX 5090 with 32 GB on an iRender server.
9. Why does Blender say All render layers are disabled?
Every view layer in the scene has Use for Rendering turned off, so there is nothing to render. Open View Layer Properties and tick Use for Rendering on at least one view layer, then render again.
Render your Blender projects on iRender
iRender is an IaaS GPU render farm: you rent a bare-metal server with 1, 2, 4 or 8 NVIDIA RTX 4090 (24 GB) or RTX 5090 (32 GB) GPUs, an AMD Ryzen™ Threadripper™ PRO CPU, 256 GB ECC RAM and a 2 TB NVMe SSD, connect by remote desktop and work on it like your own workstation, with full administrator access. You install Blender once; your server image is saved between sessions, so Blender and your add-ons stay installed.
Multi-GPU servers are billed per minute (1-GPU servers have a one-hour minimum), and sessions of 3 hours or more get 10% off. Our 4-GPU and 8-GPU servers are liquid-cooled; in our own measurements the GPUs hold 60–70°C under sustained render load. Elite support is available 7 days a week by hotline, live chat, Skype, WhatsApp and email.
This quick start shows how to start a server and render Blender on it:
Two Blender runs we recorded on our servers: 120 frames at 1080p on 4x RTX 5090 at 72 seconds per frame, and the same 4K scene on 4x RTX 5090 and 4x RTX 4090 at 87 and 108 seconds per frame per GPU:
New users get a 100% bonus on the first top-up (minimum $50). See plans on the Blender render farm page or download the iRender GPU app.




