Skip to content

How browser-based image conversion works

And what it can and cannot do, honestly.

Every image tool on QuickToolsruns on your own device. That is unusual enough that people reasonably assume it is marketing language for “we delete your files quickly.” It is not. Nothing is uploaded, because nothing needs to be.

The pipeline

When you add a file, three things happen, all inside the browser tab:

  1. Decode. The browser reads the file into a bitmap in memory using createImageBitmap() — the same decoder it uses for every image on every website you visit.
  2. Draw. The bitmap is painted onto an off-screen canvas. Any cropping, rotation, resizing or background flattening happens at this step.
  3. Encode.The canvas is handed to the browser’s encoder with a target format and quality, which produces the finished file.

That work runs in a Web Worker — a background thread — so a 12-megapixel photo does not freeze the page while it encodes. The result is handed back as a blob and turned into a download link. At no point does a network request carry your image.

What this buys you

  • No upload wait. Conversion starts the instant you drop the file. On a slow connection this is the difference between seconds and minutes.
  • No file size cap.Server-based converters impose limits because bandwidth and storage cost them money. Ours are bounded only by your device’s memory.
  • Genuine privacy. A confidential document or a personal photo never leaves your machine, so there is no copy for us to leak, subpoena, or leave in a misconfigured bucket. We could not hand it over if we were asked.
  • It works offline. Once the page has loaded, the tool keeps working with no connection at all.

How to verify that yourself

Do not take our word for it. Open any conversion tool, let the page finish loading, then turn off your Wi-Fi or put the device in airplane mode. Convert a file. It works. If the image were being sent anywhere, it could not.

For a stricter check, open your browser’s developer tools, switch to the Network tab, and convert a file while watching it. You will see no request carrying your image, because there is none to see.

What it costs

Running in a browser is not free of trade-offs, and it would be dishonest to list only the upsides.

  • One frame only. A canvas holds a single frame, so animation does not survive any conversion here. An animated GIF becomes a still image, and that is true even when the target format supports animation — writing an animated WebP needs a multi-frame encoder, which is a different piece of software from the one the browser gives us.
  • Metadata is dropped. EXIF data — camera model, settings, and often GPS coordinates — does not survive the trip through a canvas. For most people this is a feature: converting a photo before posting it strips the location it was taken. If you need the metadata preserved, this is the wrong tool.
  • The browser owns the encoder. We ask for a format and a quality level; the browser decides exactly how to encode it. That means results differ slightly between Chrome, Firefox and Safari, and it means we can only offer formats your browser can actually write.
  • Very large files depend on your device. A 100-megapixel scan may exhaust memory on an older phone. A server would not care; your laptop does.

Which formats work, and why some do not

Browsers decode far more formats than they encode. Every current browser can read PNG, JPEG, WebP, AVIF, GIF, BMP, SVG and ICO, but the canvas can only write PNG, JPEG and WebP. That asymmetry is why conversions into those three are available today and conversions into AVIF, GIF, BMP or ICO are not yet.

Two formats cannot even be read: HEIC, because it uses the patent-encumbered HEVC codec that Chrome and Firefox declined to license, and TIFF, which no browser has ever supported. Both need a WebAssembly codec shipped alongside the page — real work that is planned rather than pretended.

When a server is genuinely the right answer

Some jobs cannot be done this way and we will not claim otherwise. Video transcoding, OCR on a large scanned document, and anything requiring a lookup against an external service all need a server. Where a tool on this site needs one, its page says so plainly and describes what gets sent.

Ready to try it? Start with the image tools, or read more about how this site is built and funded.