Make a square image

A square image has two honest answers, not one

Every tool that squares an image has to decide what to do with the difference between width and height, and there are only two possibilities: throw pixels away, or add new ones. Most tools pick one silently. This page shows both, with the cost of each measured against your specific file, before anything runs.

Crop takes the short side. A 4000 by 3000 photograph becomes 3000 by 3000 and a quarter of the image falls away — the left and right edges, since the crop is centered. Pad takes the long side. The same photograph becomes 4000 by 4000, every original pixel survives, and a quarter of the result is new background above and below. Neither is correct in general. Which one is correct depends entirely on what is in the picture.

When to crop and when to pad

Crop a photograph whose subject sits near the middle. Faces, food, products and landscapes usually tolerate losing their edges, and a cropped result fills the frame with actual image rather than empty space. If the subject is off-center the centered crop will cut it, so check the preview before downloading.

Pad a logo, an icon, a diagram or a screenshot. These are the cases where losing edge pixels destroys the thing itself: a wordmark with its last letter cropped is not a smaller wordmark, it is a mistake. Padding with transparency keeps the artwork intact and lets whatever sits behind it show through. Padding with a color is the right choice when the destination will place your image on an unknown background and you would rather choose the surround yourself than let the platform pick.

Why a converted SVG so often needs this page

Vector artwork rarely has a square bounding box. When an SVG is rasterized the resulting PNG inherits whatever proportions the drawing had, and that file is then rejected or badly cropped by anything that wants a square: avatar fields, app icons, profile pictures on most networks. The fix is not to redraw the logo. It is to pad the PNG to a square with transparency, which leaves the artwork untouched and gives the platform the aspect ratio it insists on.

If your artwork is still a vector, convert it first with SVG to PNG at the resolution you want, then square the result here. Doing it in that order means the rasterizer works from the vector at full quality instead of scaling up a small bitmap afterwards.

The side is yours to choose

If you are not sure what you are starting from, check the image dimensions first — it reports width, height and aspect ratio without changing anything, and it applies EXIF rotation, so the numbers match what you actually see rather than what the file header stores.

If a square is not actually what you need — if the target is 1200 by 630, say — then resizing the image dimensions is the page for it: it scales and crops in one pass, so the picture is read from the original once instead of being resampled at each step.

1:1 is one of several shapes Instagram will keep rather than the only one, which is worth knowing before you crop a photograph to a square out of habit. Their help page keeps anything from 1.91:1 through 3:4 at its own proportions, so a portrait frame often survives better as a portrait; Instagram post size quotes the band and the pixel pairs it works out to.

The default is the short side for cropping and the long side for padding, because those are the two sizes that involve no resampling at all — the pixels are copied, not recalculated. That default is a shortcut, not a ceiling. Type any side you want and you will get exactly that, including a side larger than the original.

Scaling up does not invent detail that the camera never captured, so an upscaled result is softer than the original and the page says so when it applies. That is still frequently the right trade: a slightly soft image at the dimensions a platform demands is more useful than a sharp one it refuses. Common avatar and post sizes are one tap away, and the exact output dimensions appear on the receipt after every download, because the dimensions are the entire promise this tool makes.

Transparency survives, unless the format cannot carry it

PNG stores an alpha channel, so transparent padding downloads as genuinely transparent and the checkerboard in the preview is what you will get. JPEG has no alpha channel at all. Choosing JPEG with transparent padding means the padding has to become a solid color, and the page tells you it will be white rather than letting you discover it after upload. If you want a colored surround in a JPEG, set the background color explicitly and the preview and the download will match.

An animated GIF or WebP opens as its first frame, because a canvas is the encoder here and a canvas holds one frame. The download is a still image. That is a real limitation rather than an oversight, and it is stated here instead of being discovered later.

If the destination shows it as a circle

Avatars on most networks end up round, and that changes what a good square looks like: anything within roughly a fifth of the frame, measured from each corner inward, will be hidden by the mask the platform applies. Keep faces and wordmarks away from those four regions and a square file is all you need to supply.

Occasionally you want the round shape baked into the file itself — a signature block, a slide, somewhere that will not do the masking for you. Circle crop an image does that, and its main decision is what to leave in the corners, since a transparent PNG and a solid-colored one behave very differently once uploaded.

One image here, many in the batch workbench

This page handles one file and shows a large preview of it, which is what you want when the decision — crop or pad — needs your eyes on the specific image. When the decision is already made and the same treatment applies to a folder of files, the bulk image resizer runs the identical placement engine across a whole batch and returns a ZIP. The two share their math precisely so a single file and a batch of a thousand cannot disagree about where the crop lands.

If your problem is file size rather than shape, this is the wrong page: squaring changes dimensions, not bytes. Compress an image to a size works backwards from a target you name and shows every attempt it made.

Where the subject sits in the square

Both routes to a square need the same second decision, and it is easy to miss: which part of the image ends up in it. When you crop, the frame decides what survives. When you pad, nothing is thrown away, but the picture still has to sit somewhere inside the new square — and a logo that belongs against the top edge is as reasonable a request as a face in the middle. Drag the subject to place it, or use the sliders underneath. Padding keeps its own slider range, because shrinking the subject there is a real, visible change: it simply leaves more background around it.

Nothing is uploaded

Your browser decodes the file, draws it onto a canvas and encodes the result. No part of the image is sent anywhere, there is no account, and closing the tab is the whole of the cleanup. You can confirm that independently — open your browser's network panel and watch it stay empty while you work, or read how to verify local processing.