Upscale Image

This tool makes an image bigger in pixel dimensions. Enlarge every picture by the same factor — 2×, 3× or 4× — or set a target long side and let each image's aspect ratio decide the rest. It is for a print that needs more pixels than the camera gave you, an upload form with a minimum size, a wallpaper, a photo that has to fill a large screen, or a small logo that has to be delivered at a usable size. Up to 50 images can be enlarged in one go, and it all happens in your browser: the files are never uploaded to a server.

It cannot invent detail, and this is the most important thing to understand before using it. Enlarging with interpolation reads the pixels that are already there and estimates the new ones between them. The picture gets more pixels, edges get smoother, and with the optional sharpen pass the result looks crisper — but no new sharpness appears, because no new information was captured. A tightly cropped, blurry, or heavily compressed photo stays blurred when enlarged; it just becomes a bigger blurred photo. Recovering detail that was never recorded — reading a face or a number plate out of a soft crop — is a different, AI class of tool that guesses plausible texture rather than restoring the truth. The only way to get a genuinely sharper big picture is to start from a bigger original.

The two ways to set the size both keep the shape of the picture exactly; nothing is cropped or stretched. By factor, both sides are multiplied: a 1000 × 500 image at 2× becomes 2000 × 1000, at 3× it becomes 3000 × 1500, and at 4× it becomes 4000 × 2000. By target long side, the long side is set to the exact number of pixels you type and the short side follows the aspect ratio, rounded to the nearest whole pixel and never below 1: that same 1000 × 500 image with a target of 3000 px comes out 3000 × 1500. If the target is shorter than an image's own long side, that image comes out smaller rather than larger — the result says so, and the Image Resizer is the tool built for shrinking.

How it works: your browser decodes each image (a phone photo is turned upright using its EXIF orientation), draws it onto a canvas at the new size, and encodes the result. The drawing uses the canvas's highest-quality smoothing (imageSmoothingQuality = "high"), which is the browser's best interpolation filter rather than nearest-neighbour pixels — which filter an engine actually uses is up to the browser, but the pixels are estimated either way. If you leave "Sharpen after enlarging" ticked, a mild unsharp mask then runs over the result: each colour channel becomes its value plus a fraction of the difference between that value and the average of its 3 × 3 neighbourhood. That puts edge contrast back so the enlargement does not look soft. It is contrast, not recovered detail, and transparency is never touched.

"Same as input" keeps PNG as PNG, JPG as JPG and WebP as WebP. GIF, BMP and AVIF are saved as PNG, because browsers can read those formats but cannot reliably write them from a canvas; an animated file keeps only its first frame. You can also pick JPG, PNG or WebP explicitly, and JPG and WebP use the quality slider (92 by default — higher than the resizer's, because every pixel here is an estimate). JPG cannot hold transparency, so transparent areas are filled with the colour you choose.

Limits: each file can be up to 50 MB, and an output can be at most 16,384 pixels on a side and 100 megapixels, which is where browser tabs start running out of memory. iPhones and iPads limit a canvas to about 16.7 megapixels, so a very large enlargement may fail there with a message saying so. Enlarging always makes the file bigger, because it contains more pixels: a 2 MB photo at 4× has sixteen times the pixels, so saving it as PNG can produce a file many times the size. EXIF metadata (camera, GPS location, date) and embedded colour profiles are not copied into the result.

If you only need to change the pixel dimensions — including making a picture smaller — the Image Resizer does that with percentage, width, height and exact-box modes. To make the file smaller rather than the picture, use the Image Compressor. To change the number a printer reads without touching a single pixel, use the Image DPI Changer.

Your images stay on your device. They are read into the page's memory, enlarged there and offered back as downloads. You can confirm it in your browser's developer tools: the Network tab shows no request carrying your files.

1. The file's size is checked (50 MB) before it is read, and its first bytes confirm it really is PNG, JPG, WebP, GIF, BMP or AVIF, whatever its extension says. 2. The browser decodes it (applying EXIF orientation, so phone photos come out upright) to get its true width w and height h. 3. The new size is computed, always keeping the aspect ratio: • Factor f (2, 3 or 4): (w × f, h × f). • Target long side T: scale = T ÷ max(w, h); the long side is then pinned to exactly T and the short side is the other dimension × scale, rounded to the nearest whole pixel with a minimum of 1. 4. The result is checked against the canvas limit — 16,384 px per side and 100 megapixels — and rejected with a message if it is over. 5. The image is drawn onto a canvas at the new size with imageSmoothingEnabled = true and imageSmoothingQuality = "high", the browser's best interpolation filter. For JPG output a background colour is painted first. 6. If sharpening is on, each RGB channel is replaced with value + (strength ÷ 100) × (value − local average), where the local average is the mean of the 3 × 3 neighbourhood with edge pixels clamped at the border and the result clamped to 0–255. Alpha is never changed. 7. The canvas is encoded as PNG, JPG or WebP and named after the original with the enlargement appended, e.g. holiday-2x.jpg or scan-4000px.png.