Pixels, resampling, and why upscaling always loses

A digital image is a grid of pixels — tiny coloured squares arranged in rows and columns. A 1920×1080 image has 1,920 columns and 1,080 rows, giving you about two million pixels total. When you resize it, the software has to produce a different grid — and that means either throwing pixels away or inventing new ones.

Throwing pixels away (shrinking) is fine. You have more data than you need and the software picks the best ones to keep. The result looks good. Inventing pixels (enlarging) is the problem — the software has to guess what colour the new pixels should be based on the ones around them. This is called interpolation, and no matter how smart the algorithm, it is still a guess.

What "maintain aspect ratio" is actually doing

If your original image is 1200×800 pixels, the ratio of width to height is 1200÷800 = 1.5. When you resize to 600 wide, the tool calculates 600÷1.5 = 400 for the height automatically. Turn off aspect ratio lock and enter mismatched dimensions — say, 600×600 — and the software stretches every pixel horizontally or squashes every pixel vertically. Faces get wide, circles become ovals. It's rarely what you want.

JPG vs PNG vs WebP — an honest comparison

These three formats account for almost every image on the web, and people mix them up constantly. The differences come down to one thing: how each format decides to store colour information.

JPEG works by throwing away colour information your eyes are unlikely to notice, mainly fine detail in areas of even colour. This is called lossy compression. A fresh photograph saved as a high-quality JPEG looks almost identical to the original — but if you open that JPEG, edit it slightly, and save it again, you apply another round of lossy compression on top of the first. Do that five times and it starts looking muddy. JPEG has no transparency support, so you cannot have a subject on a transparent background.

PNG throws nothing away. It compresses images using the same lossless algorithm ZIP files use — the file gets smaller but every pixel is stored exactly as it was. PNG files are bigger than JPEG for photographs, sometimes two or three times bigger, but for anything with sharp edges, flat areas of colour, or text — logos, screenshots, diagrams — PNG looks noticeably better. It also supports a full transparency channel, which is why logo files are almost always PNG.

WebP was developed by Google and offers both lossy and lossless compression in one format. A WebP file is typically 25–35% smaller than a JPEG at the same visible quality, and it supports transparency too. The catch used to be browser support — older Safari versions did not support it — but as of 2023 all major browsers handle WebP, so there is no good reason not to use it for web images.

EXIF data, DPI, and the stuff hidden inside your images

JPEG files store more than just pixel data. Embedded inside every photo your phone takes is a block of metadata called EXIF — short for Exchangeable Image File Format. It records the camera or phone model, the focal length, shutter speed, ISO, the date and time, and if your location settings were on, your GPS coordinates to within a few metres.

This is useful when you're organising your own photos. It becomes a privacy issue when you share them. Posting a photo with GPS coordinates still embedded tells anyone who checks — and the tools to do so are freely available — exactly where the photo was taken. The "Preserve EXIF metadata" option in the tool below lets you decide: keep it for archival purposes, strip it before sharing publicly.

DPI is not what most people think it is

DPI stands for dots per inch, and it describes how many pixels are packed into each inch when printed. It has no effect on how an image looks on a screen — screens display every pixel at whatever size the screen renders it, regardless of the DPI value stored in the file.

Where DPI matters: a 3000×2000 pixel image saved at 300 DPI will print at exactly 10×6.67 inches. Change the DPI to 72 without changing the pixel dimensions and the print comes out over 40 inches wide. Change the DPI to 300 without changing the pixel dimensions and nothing changes on screen, but the printer knows to pack pixels tightly.

Things worth knowing before you resize

• Shrinking always looks better than enlarging. When you make an image smaller, the tool picks the best pixels to keep. When you make it bigger, it has to invent pixels that were never there. Start from the largest original you have — don't work from a small compressed copy.
• Never save over a JPEG repeatedly. Each time you save a JPEG, the compression runs again on top of the previous compression. The quality difference is small at first but after several re-saves you get visible blurring around edges. Keep a PNG or TIFF master and export to JPG only for the final delivery.
• For A4 at 300 DPI you need at least 2480×3508 pixels. If you don't have that resolution to start with, upscaling won't create the missing detail — the print will still look soft. The pixel count needs to be real, not invented by an algorithm.
• Use WebP with "Optimise for web" on for the smallest possible web images. Together they cut file sizes dramatically compared to PNG or unoptimised JPEG. Every kilobyte saved loads your page slightly faster.
• Turn off EXIF before sharing photos publicly. If you took the photo on your phone with location enabled, the EXIF data contains your GPS coordinates. Strip it before uploading to social media or sending to people you don't know.

Questions people usually have