Drag a crop box on the picture, or type an exact size, a width, a percentage or a target file size. The result is drawn beside it as you go, at its real dimensions and its real weight, before you save anything.
Drop images here, or click to choose
PNG, JPG, WebP, AVIF, GIF, BMP, SVG. As many at once as you like. HEIC opens in Safari only.
Free · no sign-up · nothing uploaded. There is no 10 MB cap and no file count, because there is no server paying for either. Your photos are cropped and scaled on your own machine.
One file or a folder's worth. There is no upload, so a 40 MB camera JPEG opens as fast as a thumbnail.
Pull the handles to keep just the part you want, then set an exact box, a width, a longest side, a percentage or a target file size in KB. Both write to the same result.
The panel beside the picture is the real encoded result, at its real dimensions and weight, updating as you drag. Save one file, or take the batch as a ZIP.
Most resizers give you a width field and a percentage box and stop there. Those two cover maybe half of the reasons anyone resizes a picture. The mode buttons are the other four.
| Mode | You give it | Reach for it when |
|---|---|---|
| Exact size | A width and a height | A platform has told you the pixels. This is the only mode where the picture's shape and the box's shape can disagree, so it is the only one with a fit setting. |
| Width | One number | Images for a web page or a blog post, where the column width is fixed and the height is whatever it is. |
| Height | One number | A row of logos or headshots that have to line up along a shared baseline. |
| Longest side | One number | A mixed batch of portrait and landscape photos. Every one comes back within the same bounding box, whichever way up it is. |
| Percent | A proportion | You want it smaller and you do not care exactly how small. Halving is the safest single choice. |
| File size | A number of KB | An upload form has a limit. This is the mode that answers "must be under 2 MB" directly instead of leaving you to guess at a width. |
Drag any of the eight handles to change the crop, drag inside it to move it, or drag on a part of the picture outside it to start a new one. Double-click resets it to the whole image. The box has a focus ring and takes the arrow keys too: one pixel a press, ten with Shift held, and hold Alt to resize instead of move.
The live figure inside the box is the crop in source pixels, and the panel on the right is the actual encoded file, not a mock-up. It is re-encoded as you let go of the handle, so the dimensions and the weight beside it are the dimensions and the weight of the thing you are about to download. Most resizers make you export before they tell you either.
Crop is per file and the rest of the settings are shared. In a batch, dragging the crop on one photo leaves the others alone, while a change of width or format applies to all of them. That is nearly always what you want: the size requirement comes from wherever the pictures are going, and the framing is a decision about each individual photograph.
Taking a 5926px photo down to 150px asks the browser to build each output pixel from a block of about forty by forty input pixels. It does not sample that many. It reads a smaller neighbourhood and discards the rest, which is fast, and which throws away fine texture.
So past a reduction of about four times, the scaling here halves the picture repeatedly instead and only makes the last step directly. Every step then stays inside the range the browser's filter actually covers. Below four times it does not step, because it measurably makes no difference: a 4032px phone photo taken to 1200px scored 0.9952 one way and 0.9950 the other against an exact reference, which is a tie.
Stepping is not free on every picture, and the honest version is on the guide, along with the case where it costs a little rather than gains, and the method and dates for all of it.
Resizing changes how many pixels there are. Compressing changes how many bytes each pixel costs. Both make the file smaller and they are not interchangeable.
If a picture is going to be shown at 800px wide, sending a 4000px version of it is waste no amount of compression fixes: the browser downloads four million pixels it will never draw. Resize first. If the picture is already the right size on screen and the file is still too heavy, that is a compression problem, and the quality slider here is the lever. The File size mode does both at once, which is why it asks for a maximum width as well as a budget.
It will not put back detail that is not there. Enlarging spreads the information already in the file across more pixels. It does not invent any, and no tool that runs without a model can. There is no upscaling AI here.
It will not keep your metadata. Scaling means decoding to pixels and encoding again, and EXIF, GPS and the camera settings do not survive that. The rotation does, because it is applied to the pixels deliberately.
It will not open a HEIC outside Safari. No browser except Safari can decode the format, and there is no server here to hand the file to instead. On Chrome or Firefox it says so rather than failing quietly, and the HEIC guide has the fix.
It will not rotate or straighten. A photo that needs levelling by two degrees needs a different tool. The rotation a phone records in its EXIF tag is applied on the way in, so files arrive upright, but there is no dial to turn.
No to both, and that is a consequence of the architecture rather than a generous free tier. The resizing happens in your own tab, so there is no upload bandwidth to pay for and no storage to rent. The tools that cap you at 10 MB or twenty files are paying for both. The one real limit here is your own machine: a picture has to exist as raw pixels to be scaled, at four bytes each, so anything above 100 megapixels is refused with a message rather than left to crash the tab. A 40 MB camera JPEG is nothing. A gigapixel panorama will not open.
They decide what happens when your picture is not the shape of the box you asked for, and there are only three honest answers. Fit inside keeps the whole picture and lets the result come out smaller than the box on one edge. Crop to fill fills the box exactly and loses the long edges. Pad fills the box exactly, keeps the whole picture, and puts your chosen colour in the gap. Stretch is the fourth option and it distorts the picture, which is why it is last on the row. A 4:3 photo going into a 1080×1080 Instagram square has to give up something, and this is you choosing which.
The crop happens first, and everything else is measured against what it left. Keep a 4032×3024 photo but crop it to a 2000×2000 square, then ask for a width of 800, and you get 800×800. The width you type always means the width of the part you kept, never the width of the file you started with. That is why the result panel changes as you drag even when you have not touched a number field. If you want the crop itself to hold a shape while you drag, the crop ratio buttons under the picture lock it, and Output locks it to whatever box you have set above.
Yes. Pick File size, type the number, and set a maximum width to scale within. The tool then encodes the picture several times at different quality settings, narrowing in on the highest quality that still fits your budget, and tells you which quality it landed on. If the target cannot be reached at that width it says so rather than handing back an oversized file and calling it done. That happens when the budget is very small and the width is large: 20 KB from a 1920px photo is a request no encoder can meet, and lowering the width is the fix.
It will if you turn "never enlarge" off, and mostly it is not worth doing. Enlarging cannot recover detail that is not in the file, so a 400px image asked for at 1600px gets four times as many pixels carrying the same amount of information. The result is soft, and on hard edges it is soft with a halo. The honest uses are real but narrow: hitting a platform's stated minimum, or padding a small logo out to a required canvas. For the second, leave "never enlarge" on and choose Pad. The picture stays sharp at its own size and the canvas comes out the size you asked for.
Because the format changed, almost always from JPG to PNG. PNG throws nothing away, so a photograph written as one can easily be several times the size of the JPG it came from, even at a quarter of the dimensions. The result panel shows the new file size before you save anything, so you can see it happen and change the format on the spot. The other cause is a very small reduction on an already-heavily-compressed JPG: re-encoding at quality 85 something that was saved at 60 will add bytes back. It reports the change either way rather than assuming a resize made things smaller.
Yes, all of it, and it cannot be turned off. The picture is decoded to pixels, scaled, and encoded again, and nothing that was not a pixel survives that. The camera body, the lens, the timestamp and the GPS coordinates are all gone from the file you save. For a photo going on the public web that is usually the outcome you wanted. The rotation is the exception, and it is handled deliberately: a sideways phone photo has its orientation baked into the pixels on the way through, so the file you get is upright on its own.