The Background Removal Shakedown

You need to remove the background from a photo. Product shot for your online store, headshot for LinkedIn, flyer for your kid's soccer team. You search for "remove background from image" and click the first result.

You upload your photo. Three seconds later, the background is gone. The result is perfect. Clean edges, natural hair detail, exactly what you needed. You right-click to save it.

The image is 625 by 400 pixels. Tiny. Unusable. Your original was 4000 by 3000. The tool processed the full-resolution image, showed you it could do exactly what you wanted, and gave you a thumbnail.

Full resolution: $1.99. Or subscribe for $9 to $99 per month.

Show, then ransom

This model is different from the other traps. QR code generators deactivate your codes after a trial. Resume builders paywall the download after you've typed for an hour. PDF tools lock the save button.

Background removal tools do something sharper. They show you the finished product. Not a preview. Not a mockup. The actual result, at full quality, right there on your screen. Then they shrink it.

remove.bg

remove.bg, originally built by Vienna-based Kaleido AI and acquired by Canva, is the biggest name in this space. 60 million monthly visits. The free tier processes your image at full resolution and delivers a 625x400 pixel download. About 3% of the pixels in a typical smartphone photo.

A small e-commerce seller photographing 50 products would pay $99/month. $1,188/year for an operation that now runs entirely in a browser tab.

PhotoRoom

PhotoRoom raised $62 million, hit $94.5 million in annual recurring revenue (89% year-over-year growth), crossed 100 million app downloads, and reached a $500 million valuation.

In 2024, PhotoRoom did something that went beyond aggressive pricing into breaking trust with paying customers. Users who had prepaid for annual "Pro" subscriptions, which explicitly included unlimited batch exports, found their accounts capped at 500 batch exports per month mid-contract. PhotoRoom introduced a new, more expensive "Ultra" tier and moved the features people had already paid for into it.

They refused refunds. Downgrading a paid user's existing plan isn't a price increase. It's taking back something someone already bought.

What changed

Two years ago, charging for background removal was defensible. The AI models were big, needed GPUs, and couldn't run in a browser. Server costs were real. That changed fast.

ONNX Runtime Web brought neural network inference to the browser through WebAssembly and WebGPU. Models that needed a server with a GPU now run on the user's device. WebGPU + WebAssembly enables up to 20x speed improvement over CPU-only browser inference. Background removal is effectively instant after the initial model download (~30MB, cached). Not just possible in the browser — it's faster than uploading to their servers and waiting for a response.

IS-Net and BiRefNet are open-source segmentation models (Apache 2.0 and MIT licensed) that produce commercial-quality results. These aren't research demos. They match remove.bg's output quality.

@imgly/background-removal-js (6,000+ GitHub stars) packages these models into a JavaScript library that runs entirely in the browser. Full resolution, no upload, no server.

The computation that remove.bg charges $1.99 per image for costs $0.00 to run in your browser.

Who gets hurt

E-commerce sellers. A seller listing 200 products needs background-free images for every listing. At $1.99 per image, that's $398 for one batch. Small sellers on thin margins absorb this because they don't know there's an alternative.

Freelancers. Designers and photographers doing background removal for client work. A freelancer processing 20 images per week pays over $2,000/year at remove.bg's per-image rate.

Anyone who already paid for PhotoRoom. The users who committed to annual plans got hit hardest. Prepaying is supposed to be a discount in exchange for commitment. When the company downgrades your plan after payment, it penalizes the people who trusted them the most.

The free alternative

I built nah.tools/photo because there's no reason to pay per image for something your browser does for free.

The background remover processes images entirely in your browser using open-source AI models. Your photos never go to any server. Full resolution output, same pixels as your original minus the background. No watermark. No resolution gate. No per-image fee.

First use downloads the AI model (about 40MB), which your browser caches automatically. Every use after that is instant. Source code is public. Verify that no image data leaves your browser.

How to protect yourself

1. Check output resolution before uploading sensitive images. If a tool processes your photo but delivers a shrunken version, the restriction is artificial.
2. Prefer tools that process locally. When you upload to a server-based tool, your image is on someone else's computer.
3. Watch for mid-contract changes. If you're on an annual plan, screenshot the feature list when you subscribe.
4. For batch processing, free tools exist. rembg (16,000+ GitHub stars) handles unlimited images locally in Python.

A background removal tool that shows you the perfect result and then asks you to pay for the pixels is not offering a free trial. It's showing you what it's withholding.

The AI is open source. The runtime is open source. The computation happens on your device. The only thing between you and your image is a business model built on artificial scarcity.