Resize Images Online

Resize images to exact dimensions

Photos straight off a camera or phone are often far bigger than any page or platform needs. This resizer lets you set precise pixel dimensions or scale by percentage, with an aspect-ratio lock so nothing stretches — and it does the whole job in your browser with a live before-and-after preview.

How resizing works

Your image is drawn onto an HTML canvas at the target width and height, then re-encoded in the format you choose. With the aspect-ratio lock engaged, the tool calculates the matching dimension for you, so a 4000×3000 photo scaled to 800 wide automatically becomes 800×600. Quick-scale buttons apply common percentages in a single click.

Interpolation: how new pixels are decided

When a target dimension does not line up with the source grid, the resampler has to invent the in-between values. The method it uses to do that is interpolation, and the choice is what separates a crisp result from a blocky or mushy one.

• Nearest-neighbour simply copies the closest existing pixel. It is the fastest and the only method that keeps hard pixel boundaries intact, which is exactly why it is right for pixel art and screenshots of low-resolution UI — and exactly why it produces jagged staircases on photographs.
• Bilinear blends the four nearest pixels by distance. It is smooth and cheap, but on large reductions it can look a touch soft because it samples too small a neighbourhood to represent the detail being discarded.
• Bicubic weighs a 4×4 block with a cubic curve, preserving edge contrast better than bilinear. It is the default behind most photo software and the reasonable middle ground for everyday resizing.
• Lanczos uses a windowed-sinc kernel over an even wider neighbourhood, giving the sharpest downscales with the least aliasing — at the cost of occasional faint ringing near very high-contrast edges.

This resizer asks the canvas for its highest smoothing quality, which on Chrome and modern Firefox is a Lanczos-style filter and on Safari a high-quality bicubic. In practice that means downscales come out clean without you choosing a kernel by hand. The one case where you would want nearest-neighbour instead — sprite sheets and pixel art — is better handled by an editor that exposes that mode explicitly.

When the algorithm matters most

The interpolation choice is barely visible when you shrink a photo by 10–20%. It becomes obvious in two situations: aggressive downscaling, where a 4000-pixel image collapses to 400 and a weak filter throws away detail unevenly, producing moiré on fine repeating patterns like fabric or brick; and any upscale, where every method is guessing and the differences in softness are plain to see. As a worked example, a 3000×2000 photo reduced to 600×400 keeps legible text on a sign with a Lanczos-style filter but smears it with plain bilinear — a four-fold dimension drop is right at the boundary where the kernel starts to count.

Common target sizes

Use 1080×1080 for a square social post, 1920×1080 for a full-HD banner or slide, 1280×720 for a YouTube thumbnail, and around 1600–2048 px on the long edge for crisp blog images. When a form or avatar field caps the dimensions, type the exact limit and let the lock handle the other side. If a resized file comes out larger than expected, lower the JPG/WebP quality or follow up with the compressor.

Resized privately in your browser

The resampling happens on a canvas the page draws in your own tab, so the picture is redrawn at its new size and handed straight back for download without a single byte being sent to us.

Related image tools

• Smart Image Upscaler — going bigger, not smaller? Multi-pass resampling beats a plain stretch.
• Bulk Image Resizer — same long-edge limit across a whole folder in one pass.
• Image Cropper— fix the framing or aspect ratio that resizing alone can't.
• Image Compressor— drive the byte count down after you've fixed the pixels.