Choose a photo to resize it.
A phone held sideways stores the photo sideways, with a small tag saying which way up it goes.
Every viewer honours that tag. A canvas does not, unless it is told to — and a canvas export carries no tag, so once the photo has been redrawn the information is gone and it stays rotated forever. That is why resized photos come out sideways in almost every tool on the web.
This page reads the tag with the Metadata Remover’s own parser and turns the pixels themselves before saving. One place knows which way up the picture goes.
And there are two things no tool can give you.
Enlarging adds no detail. It fills in between the pixels that exist, so what you get is a smoother, softer version of the same picture.
And re-encoding a JPEG loses something every time, even at quality 100 — which means “quantise as little as possible”, not “do not quantise”. It is why the Metadata Remover beside this page uses no canvas at all.
Because an iPhone writes HEIC by default, and no browser except Safari on Apple hardware can decode one — the codec belongs to the operating system, not to the browser. Reading it here would mean downloading several megabytes of decoder before your photo could even be opened, which this page has not taken on. Two ways round it: set Camera to Most Compatible in Settings so the phone writes JPEG, or share the photo out of Photos, which converts it on the way. Either way what you get is a JPEG and this page will handle it.
No. The file is read from your disk into this browser tab, resized by the browser itself, and handed back as a download. There is no upload and no server — you can watch the network tab while you use it. Every other resizer that asks you to upload first has your photo on somebody else’s disk before it does anything.
A phone held sideways stores the photo sideways along with a tag saying which way up it goes. Viewers honour the tag; a canvas does not unless it is told to — and a canvas export has no tag, so once the photo is redrawn the information is gone and it stays rotated. This page reads the tag itself and turns the pixels the right way up before saving, so the result is upright permanently.
No, and no tool can. Enlarging fills in between the pixels that exist; it cannot recover detail that was never captured. What you get is a smoother, softer version of the same picture at a larger size. The page says so when you ask for one, rather than quietly producing it.
Usually one of two reasons. Saving a photograph as PNG makes it several times larger — PNG is lossless and stores photographs very inefficiently, and it is for flat colour and sharp edges. Or the quality is set above what the original was encoded at, in which case the encoder spends more bytes preserving the compression artefacts that were already there.
It encodes the image several times at different quality settings and keeps the best one that fits, narrowing the range each round. There is no formula for it: the same quality that gives 180 KB for a portrait gives 900 KB for a photo of leaves, so the only honest way to hit a target is to try. It stops after eight attempts and tells you what quality it settled on.
It says so, and gives you the smallest file it could make. Past a point a picture cannot be made smaller by throwing away more detail — the quality is already at the floor and what remains is the sheer number of pixels. Reduce the dimensions as well and it will fit.
That is JPEG, which has no transparency at all — every see-through pixel has to become something. This page paints it white rather than leaving it black, which is what most people expect, and warns you before you download when the image actually has transparency in it. Save it as PNG or WebP to keep it.
Yes, always, even at quality 100 — that setting means “quantise as little as possible”, not “do not quantise”. The blocks are decoded, resampled and quantised again, and every round trip softens it a little. If all you wanted was to strip the metadata, the Image Metadata Remover does that without re-encoding anything.
Two, and both are about memory rather than policy. The file itself is capped where the page says, and the output is capped at 20,000 pixels on an edge — a canvas needs about four bytes per pixel while it is being drawn, so a 48-megapixel image is roughly 190 MB before a single byte is encoded. That is most of what a phone browser will give a tab.
There is no formula. The same quality that gives 180 KB for a portrait gives 900 KB for a photo of leaves — more detail, more bytes. So it encodes repeatedly, narrowing the range each round, up to eight attempts. The only honest way to hit a target is to try.
The photo is read and redrawn in your browser and nothing about it is transmitted or stored. Resizing re-encodes the image, so some quality is always lost — and the result carries no metadata at all, including the orientation tag, which is why the pixels are turned the right way up before saving. Nothing you type here is sent to our servers — the calculation runs entirely in your browser.
