Image resizer

How the resizing is done

Naively scaling a large image to a small one in a single step produces aliasing: the browser samples widely spaced pixels and drops everything between them, which shows as jagged edges and shimmering detail on anything with fine texture.

This tool downscales in halving steps instead. Each pass reduces the image by at most half, which is effectively a box filter that averages neighbouring pixels, and repeats until it reaches the target. The last step lands exactly on the requested dimensions. The result is visibly cleaner on detailed images - text, foliage, patterned fabric - than a single-pass resize.

Pixels or percentage

Exact pixels is right when a specification exists: a platform requiring 1080×1080, a CMS with a 1200px content width, a form demanding a 600×600 photograph.

Percentage is right when you want proportional reduction across images of different sizes. Scaling a mixed batch to 50% preserves their relative dimensions.

Aspect ratio

The lock keeps the original proportions: change the width and the height follows. Both are recomputed from the original ratio each time rather than from the previous value, so repeated adjustments never accumulate rounding drift.

Unlock it only when you deliberately want to stretch - or when you have already cropped to the target shape.

Common target sizes

Use	Dimensions
Instagram square post	1080 × 1080
Instagram story / Reel	1080 × 1920
YouTube thumbnail	1280 × 720
Open Graph / social preview	1200 × 630
Full HD wallpaper	1920 × 1080
Blog content image	1200–1600 wide
Email header	600 wide

Sizing for the web

The most common mistake is uploading a 4,000-pixel camera image into a layout that displays it at 800 pixels. The browser downloads every one of those pixels and then throws most of them away.

A reasonable rule: export at roughly twice the display width, to stay sharp on high-density screens, and no more. An image shown at 800 CSS pixels wants a 1,600-pixel file. Beyond that you are paying bandwidth for detail no screen will resolve.

For Indonesian, Indian and Philippine audiences on mobile data this is not a marginal concern - it is often the difference between a page that loads and one that is abandoned.

Enlarging

Scaling up interpolates between existing pixels. It can make an image fit a larger frame, but the detail is invented and the result is soft. There is no browser-side technique that recovers information the sensor never recorded.

If you need a larger image: go back to the original file, re-export from the raw or vector source, or retake it. Where that is impossible, machine-learning upscalers can hallucinate plausible detail - but that is a fundamentally different operation from resizing, and not one that belongs in a tool that promises to leave your image alone.