About
Keep the soul, drop the weight.
Ruah is a small set of image tools that do their work inside your browser — no uploads, no accounts, no limits. This is the thinking behind it.
Why “Ruah”
Ruah is an old word for breath — the living part of a thing. It's a fitting name for what these tools try to do: make an image lighter without letting the life out of it. A good compression isn't the one that shrinks a file the most; it's the one you can't tell happened. Drop the weight, keep the soul.
The idea: your computer is enough
Most image tools online work the same way — you upload your picture to a company's servers, they process it, and they hand it back. That model made sense a decade ago. It doesn't anymore. Modern browsers can run the same compression, conversion, and even AI background-removal code directly on your machine, at full speed, using WebAssembly and your device's own graphics hardware.
So Ruah doesn't have an upload button. Your file is opened, transformed, and saved without ever leaving the tab you're looking at. The server's only job is to send you the page.
What that quietly changes
- It's genuinely free and unlimited. There's no per-image server cost to pass on to you, so there's no meter, no “pro” gate on the useful features, and no daily cap.
- It's private because of how it's built, not because we promise. A tool that never receives your image can't leak, sell, or lose it. You can watch the Network tab and see for yourself.
- It's fast. Nothing waits on an upload and a download round-trip. Big batches process as quickly as your machine can go.
How it works under the hood
None of this is a trick. The everyday tools — compress, convert, resize, crop — run WebAssembly builds of the same image codecs that desktop software relies on, so the maths is identical to a native app; it just happens inside the tab. The heavier jobs use your graphics card through WebGPU, and fall back to the CPU when it isn't available.
The AI tools are real neural networks, downloaded once and then run locally. Background removal usesBiRefNet, an open segmentation model, to work out which pixels are the subject; upscaling uses Swin2SR, a super-resolution model, to reconstruct detail instead of stretching it. Both are compact on-device models, cached by your browser after the first run, so you download each one only once.
What Ruah is not
It isn't a cloud service, an account you have to manage, or a business built on your data. There's nothing to sign up for and nothing stored about you. Ads keep the lights on; that's the whole model.
The toolkit
Everything runs on-device, one drop away:
- Compress — shrink PNG and JPG, visibly lossless.
- Convert to WebP, AVIF, JPG or PNG.
- Resize and crop & rotate in batches.
- Upscale — enlarge with an AI super-resolution model.
- Remove a background with a local AI model, or erase a flat colour by hand.
- Strip metadata and EXIF without re-encoding.
Say hello
Ruah is young and still growing. If something is missing, broken, or could be better, tell us on thecontact page — it's the fastest way to shape what comes next.