Free alternative to Cleanup.pictures
Home › Guides · Last checked 2026-10-21 · Facts verified from each project's docs unless marked tested
The closest free alternative to Cleanup.pictures Pro is IOPaint, a free open-source (Apache-2.0) inpainting tool you run on your own computer. It erases unwanted objects, people and defects with the LaMa model and runs on CPU, GPU or Apple Silicon. Your images stay on your machine and there's no export size cap. The catch: you install it with Python (or a Windows installer) and open it in your browser, and the repo hasn't had a code push since April 2025.
| Cleanup.pictures | IOPaint | |
|---|---|---|
| Cost | Free, or Pro at $5/month or $36/year (Cleanup.pictures pricing) | Free |
| Export size | 720px on free; no limit on Pro | No limit set by IOPaint |
| Runs where | In your browser, on their servers | Your computer (opens in your browser at localhost) |
| Hardware | Any device | CPU, GPU or Apple Silicon |
| License | Commercial service | Apache-2.0 |
| Catch | Low-res free exports | Install and model downloads; no code push since April 2025 |
Install and run
- Install Python, then run
pip3 install iopaint. For an NVIDIA GPU, install the CUDA build of PyTorch first, per the README. - Start it with
iopaint start --model=lama --device=cpu --port=8080and open http://localhost:8080. - Models download automatically on first start.
- On Windows you can use the Windows 1-click installer instead: unzip, run win_setup.bat, then win_start.bat. Its page notes some Windows 11 machines hit an fgbemm.dll error with a linked fix.
Tips for clean results
- Brush a little wider than the object, including its shadow.
- Use LaMa for plain removals. Switch to a diffusion inpainting model only when you need to replace something.
- For product shots, pair it with our free remove.bg alternative.
The catches
- Setup: Python install or a Windows batch-file installer, not a website you drag a photo onto.
- Disk space: each model is a separate download.
- Maintenance: the GitHub repo's last code push was April 2025 and the last PyPI release (1.6.0) was March 2025, so expect fewer fixes.
When Cleanup.pictures is still worth paying for
If you clean a few photos a month on a phone or a work laptop where you can't install software, $3 to $5 a month buys high-res exports with zero setup. IOPaint wins when you edit often, want full resolution for free, or don't want to upload client photos.
Sources: IOPaint README and GitHub license (Apache-2.0), Windows 1-click installer, IOPaint on PyPI, LaMa (Apache-2.0), Cleanup.pictures pricing (all checked 2026-10-21). We have no affiliate link with either. See all free image tools or read free Topaz Gigapixel alternative.
FAQ
What does Cleanup.pictures limit on its free plan?
Its pricing page (checked 2026-10-21) says free exports are limited to 720px. Pro removes the size limit for $5 a month or $36 a year, per the same page.
Is IOPaint really free?
Yes. The README calls it completely free and open source, and the GitHub repo is licensed Apache-2.0.
Do I need a GPU?
No. The README says it supports CPU, GPU and Apple Silicon, and its quick start runs the LaMa erase model on CPU. The larger Stable Diffusion inpainting models are much slower without a GPU.
Can it do more than erase?
Yes. With diffusion models it can replace objects and outpaint (extend an image), and plugins add Segment Anything selection, background removal, Real-ESRGAN upscaling and GFPGAN face restoration, per the README.
Can I clean a whole folder of images?
Yes. The iopaint run command batch-processes a folder of images with a matching folder of masks, or one mask for all of them.
Can I remove watermarks with it?
Technically yes, but only on images you have the rights to. Cleanup.pictures' own FAQ makes the same point: watermarks usually mean the image is copyrighted.