Everyone has seen a stretched image at some point — a logo that looks weirdly squashed, a profile photo where someone's face looks slightly too wide. It's one of the most common and most avoidable image mistakes, and it comes down to a single concept that takes about thirty seconds to actually understand.
Why the "squished" look happens
Every image has an aspect ratio — the proportion between its width and its height. A 1200×800px photo has a ratio of 3:2. If you resize that image to, say, 1200×1000px without adjusting the width to match, you've just told the computer to stretch every pixel vertically to fill the extra height. Circles become ovals. Faces get subtly wider or taller than they should be. It's not corruption or a bug — it's exactly what you asked the software to do, just not what you meant.
The fix is keeping width and height locked to the same proportion whenever you resize. If you halve the width, you halve the height too. Most resizing tools have a "lock aspect ratio" option for exactly this reason — toggle it on, change one dimension, and the other adjusts automatically to keep the proportions correct. Our Image Resizer has this on by default (a "Lock aspect ratio" checkbox), specifically so the common mistake is harder to make than the correct outcome.
Downscaling vs upscaling — these are not the same operation
Making an image smaller (downscaling) almost always looks fine. You're removing pixels and the software just has to decide which ones to keep and how to blend neighbouring ones — there's more detail available than the smaller image needs, so the result looks sharp and clean.
Making an image larger than its original resolution (upscaling) is a fundamentally harder problem, and it's worth understanding why. If you take a 400×300px photo and resize it to 1600×1200px, the software now has to invent four times as many pixels as it started with. It doesn't have any new real detail to draw on — it can only guess, based on the colours of nearby pixels, what the "missing" pixels probably should look like. Push this too far and you get the telltale soft, blurry, slightly smudged look of an over-upscaled image. There's a hard ceiling on how much you can enlarge an image before this becomes obvious, and no amount of clever resizing settings changes that ceiling — basic interpolation can only estimate, not invent.
There's a separate, more complex category of tool — AI upscalers — that use trained models to generate plausible new detail rather than just interpolate between existing pixels, and they can produce genuinely sharper enlargements than traditional resizing. That's a meaningfully different (and much heavier) kind of processing than what a browser-based image resizer does, and it's not something you'll get from a simple resize tool — if you need real upscaling, that's a distinct category of tool to look for.
Interpolation, in plain terms
When software resizes an image, it has to decide the colour of every new pixel based on the pixels around it — that decision process is called interpolation, and different methods trade speed for smoothness.
Nearest-neighbour is the crudest method: for each new pixel, just copy the colour of whichever original pixel is closest. It's fast, but it produces visibly blocky, jagged edges — you'll notice this most on diagonal lines and curves, which end up looking stair-stepped rather than smooth.
Bilinear and bicubic interpolation are smarter: instead of just copying the nearest pixel, they blend several surrounding pixels together, weighted by distance, to produce a smoother transition. Bicubic (which considers more surrounding pixels than bilinear) generally produces the smoothest, most natural-looking result of the common methods, which is why it's the default in most quality image resizing tools, including browsers' own image rendering.
The practical takeaway
Keep aspect ratio locked unless you deliberately want a stretched effect. Downscale freely — it's basically always safe. Treat upscaling as a one-way, lossy operation with real limits, and don't expect a resize tool to add detail that was never captured in the first place. Our Image Resizer handles the locking and the interpolation for you — just set your target dimensions (or pick a preset) and it resizes correctly, entirely in your browser.