Resize an image to any size

Shrinking, yes: going smaller discards pixels you do not need at the new size, using high-quality bicubic interpolation, and the result stays sharp. Enlarging is the opposite direction: the tool has to invent pixels that were never captured, so the image softens. Always resize down from the largest original you have, and keep that original safe.

Switch the mode to Exact size and enter 1080 by 1080 for a square post, 1080 by 1350 for a portrait post, or 1080 by 1920 for a story or reel. Or just tap the Instagram preset above the steps and the tool sets itself up, including cropping to the right shape first so nothing looks stretched.

That happens in Exact size mode when the target shape does not match the original shape: forcing a wide photo into a square distorts it. Fix it by adding a Crop step before the resize so the image is cut to the right proportions first, or use Width mode, where the height follows automatically and nothing can stretch.

Switch the mode to Scale by percent and type the percentage you want, like 50 for half size. It applies to every image in a batch, which is handy when your photos have mixed sizes and you just want them all smaller.

Up to 10,000 by 10,000 pixels. Very large outputs use more browser memory, so a huge batch of huge images may be slow on an older phone, but modern devices handle them without trouble.

No. Resizing runs entirely in your browser through the HTML5 Canvas API. Your images never leave your device, and you can prove it: turn off your wifi and the tool keeps working.

Yes. Drop several images in one go and the same settings apply to each, processed one after another to keep memory stable. Download the results individually or as a single ZIP. Fit-within mode is best for mixed batches, since each image scales to fit the box without stretching.

Yes. Click Add step and add a Convert step to switch to JPG, PNG or WEBP, or a Compress step to target an exact file size like 200 KB. Everything runs in one pass on one upload, and the image is only re-encoded once at the end, which preserves more quality than saving between separate tools.