How to resize an image without losing quality

A practical guide · Updated June 2026

Resizing an image sounds like it should be one button, but the results people get vary wildly — some come out crisp and fast-loading, others end up soft, stretched, or somehow bigger than before. The difference is almost never the tool; it is understanding what "size" actually means and which direction you push it. This guide covers the handful of ideas that matter: pixels versus inches and DPI, why locking the aspect ratio saves you, why enlarging destroys quality when shrinking does not, the right dimensions for email, web, print and social, and the order to do things so your picture stays sharp. Everything here can be done for free and privately, right in your browser.

What "size" really means: pixels, not inches

The single biggest source of confusion is that an image has two different kinds of "size," and they measure completely different things. The one that matters on screen is pixel dimensions — the width and height in pixels, like 1920 × 1080. That number is the actual amount of picture information the file contains. The other is physical size in inches or centimetres, which only exists in relation to a print resolution.

DPI (dots per inch), often called PPI for screens, is simply the bridge between the two. It tells a printer how many pixels to pack into each inch of paper. Change the DPI number and you change the printed size — but you do not add or remove a single pixel, and nothing about how the image looks on a screen changes at all. A 3000 × 2400 photo is the same file whether it is labelled 72 DPI or 300 DPI. At 300 DPI it prints at 10 × 8 inches; at 150 DPI the same pixels print at 20 × 16 inches, just less densely.

The practical takeaway: for anything on a screen, ignore DPI entirely and think only in pixels. Websites, email, and social platforms measure everything in pixels and discard DPI metadata. DPI only becomes real when ink hits paper.

Locking the aspect ratio — and why it saves you

Aspect ratio is the relationship between an image's width and height — 3:2 for most camera photos, 16:9 for video frames, 1:1 for a square. When you resize, you almost always want to keep that ratio constant so the image scales evenly. Nearly every resizer has a small link or padlock icon labelled "keep aspect ratio" or "constrain proportions." Leave it on.

Here is why it matters: if you type a new width and a new height that do not match the original proportions, the software has no choice but to squash or stretch the picture to fit. Faces get fat or gaunt, circles turn into ovals, and straight buildings lean. With the ratio locked, you only set one dimension — say the width — and the other is calculated automatically so everything stays in proportion. A quick sanity check: divide the original width by the original height, then do the same for your target numbers. If the two results are not equal, something will distort.

• Lock the ratio, set one dimension. Choose the width for landscape images, the height for portraits, whichever is the limiting side.
• Need a different shape? Do not stretch — crop instead. Cropping removes pixels from the edges to change the ratio while leaving the remaining pixels undistorted.
• Watch percentages. Scaling to "50%" always preserves the ratio automatically, which makes it a safe way to halve an image.

Downscaling vs upscaling: why enlarging loses quality

This is the heart of "without losing quality." Resizing down and resizing up are not mirror images of each other — one is basically free, the other is lossy by nature.

When you downscale (make an image smaller), you are throwing away pixels you did not need. The software intelligently averages groups of pixels into fewer ones, and because a screen was never going to show all that detail anyway, the result looks just as sharp. A 4000-pixel photo shrunk to 1200 pixels for a blog post looks flawless.

When you upscale (make an image larger), there is a problem: the extra detail simply does not exist. The camera captured a certain number of pixels and no more. To fill a bigger canvas, the software has to invent pixels by guessing what belongs between the real ones. That guessing — interpolation — produces soft edges, blurry textures, and blocky artefacts. You can enlarge a little (10–20%) before it becomes obvious, but doubling the size of a small image always looks worse than the original, and the file gets bigger for detail that is not really there.

Rules of thumb worth remembering:

• Always start from the largest original you have. Never resize down, save, and then try to go big again from that smaller copy.
• Down is safe, up is not. If you must enlarge, keep it under about 25% and expect some softening.
• Small source, big need? Re-shoot, re-export, or find a higher-resolution version rather than upscaling. No tool can recover detail that was never recorded.

The right size for common jobs

There is no single "correct" size — the right dimensions depend entirely on where the image will be seen. Match the pixel width to the space it will actually occupy, and you get sharp results with the smallest possible file. Here are sensible defaults for the most common needs:

• Email attachments: 1024–1600 px on the long side is plenty to look good on any screen while staying small enough to send. Providers cap attachments around 20–25 MB, so resizing first avoids bounced messages.
• In-article web images: 800–1200 px wide. Most content columns are under 800 px, so this covers standard and high-density (Retina) screens without waste.
• Full-width web banners / hero images: 1600–2000 px wide. Beyond 2000 px you rarely see any benefit on the vast majority of displays.
• Thumbnails and avatars: 150–400 px square. Tiny slots need tiny images.
• Social posts: follow each platform's spec — 1080 × 1080 for an Instagram square, 1080 × 1350 for a portrait, 1280 × 720 for a YouTube thumbnail. Our social media image sizes guide lists them all.
• Print: here DPI matters. Aim for 300 DPI at the final print size. For a 6 × 4 inch photo that means roughly 1800 × 1200 pixels; for an A4 page, around 2480 × 3508 pixels. Below 200 DPI print starts to look soft.

A quick formula for print: multiply the inches you want by 300 to get the pixels you need. If your original does not have that many pixels, print smaller rather than upscaling.

Resample quality: the setting most people miss

When a resizer changes pixel counts it has to resample — recalculate the colour of every new pixel. The method it uses affects how clean the result looks, and better tools let you choose or use a good default automatically:

• Nearest neighbour is the fastest and worst for photos — it just copies the closest pixel, producing jagged, pixelated edges. It is only useful for pixel art you want to keep crisp.
• Bilinear averages the four surrounding pixels. It is fine and fast, but can look slightly soft.
• Bicubic looks at a wider neighbourhood and produces smooth, natural results — the standard choice for photographs and the sensible default.
• Lanczos is a high-quality method that keeps edges especially sharp when downscaling; many modern browser-based tools use it or bicubic under the hood.

For everyday resizing you rarely need to touch this — a good resizer applies bicubic or Lanczos automatically. Just know that if a shrunk image looks unexpectedly soft or jagged, the resampling method is the likely culprit.

Resize first, then compress — the order matters

Resizing and compressing are different jobs, and doing them in the right sequence gives the best-looking, smallest file. Resizing reduces the number of pixels; compression stores those pixels more efficiently at the same dimensions. Do them in this order:

1. Resize down to the pixel dimensions you actually need. This removes the bulk of the data geometrically — halving width and height cuts about 75% of the pixels before you compress anything.
2. Then compress the result at around 75–80% quality for JPG or WebP. At that level the file shrinks dramatically with no visible loss.

Doing it the other way round — compressing a giant image and then shrinking it — wastes effort, because you spend care optimising pixels you are about to throw away, and the second step can re-introduce artefacts. Resize, then compress. If you want the deeper version of this, see our guide on reducing image file size without losing quality.

How to resize an image privately in your browser

You do not need to install anything or upload your photos to a stranger's server. Imagloo's tools run entirely on your device using your browser's own image engine, so nothing ever leaves your computer — ideal for client work, personal photos, and unreleased graphics. Here is the whole routine:

1. Open the image resizer and drop in your highest-resolution original.
2. Keep the aspect ratio locked and set one dimension — usually the width — to the size the image will actually be displayed at.
3. Only scale down. If your target is larger than the original, stop and find a bigger source instead of upscaling.
4. Need a different shape? Use the image cropper first to reframe to the ratio you want, then resize to exact pixels.
5. Compress the result in the image compressor at about 75% quality, and download. The tool shows the new file size instantly so you can confirm the savings.

Follow those steps and you will get images that look identical to the original at a fraction of the size — the practical definition of resizing without losing quality.

Frequently asked questions

Try the free tool — private and in your browser

Resize any image to an exact size in seconds with Imagloo's free tools. Everything runs locally on your device, so your images are never uploaded:

• Image Resizer — scale to an exact width or height with the aspect ratio locked.
• Image Cropper — reframe to any ratio before you resize.
• Image Compressor — shrink the file after resizing while keeping quality.