Image Tools

Every image incumbent on the web works the same way: your file is uploaded to a server, processed there, and offered back — metered by file count, watermarked, or held behind a queue. The tools in this category invert that model completely. Modern browsers can decode, transform and re-encode images themselves, so every operation here runs on your own device through the canvas pipeline: the file is read locally, processed locally and downloaded from your own machine. Nothing is uploaded, because nothing needs to be — and that is not a policy promise that could change, it is the architecture of the page, verifiable by using the tools with your connection switched off.

The pipeline handles the detail phone photos are famous for: orientation. A portrait shot from a phone is usually stored as a landscape frame plus an EXIF orientation flag, and naive tools display it sideways. Every tool here reads that flag on load and corrects the picture before you see it, so what you compress, crop or sample is the upright image, and the dimensions reported are the upright dimensions. Sizes are reported with the same exactness: before and after byte counts are the real sizes of the real files, and pixel dimensions are read from the decoded image, not estimated — the compression percentage on the page is arithmetic on those two measured numbers.

The category covers the everyday jobs: shrinking a photo to an upload limit, resizing to the exact dimensions a platform demands (the preset table is printed on the Resizer page), cropping to an aspect ratio with a source-pixel readout, generating the full favicon set and its install snippet from one image, converting an image to a Base64 data URI for embedding — with the size overhead stated honestly, since Base64 costs roughly a third again — and sampling exact colours from any pixel of a picture for design work. The same pipeline carries the rest of the roster: palette extraction, which counts an image's pixels into dominant colour groups and reports each with its exact share; SVG rasterising, which renders a vector file to a PNG at the scale or exact size chosen; grid splitting, which cuts an image into tiles along rounded boundaries — no pixel lost or duplicated — and delivers them together as a ZIP; watermarking, as a single placement or tiled across the whole picture; and rotate and flip, the exact right-angle transforms whose output dimensions swap or hold by a stated rule. Eleven tools, one rule: the file is opened on your device, worked on there, and handed back there.