Image resizing is one of those tasks that seems trivial until you encounter blurry logos, stretched product photos, or profile pictures that look pixelated on retina displays. Getting it right requires understanding what happens to pixel data when you change an image's dimensions, and which settings produce the sharpest result.
How image resizing works
A digital image is a grid of pixels, each with a colour value. Resizing means either adding pixels (upscaling) or removing them (downscaling). When you downscale an image: making it smaller: the software must combine multiple original pixels into fewer output pixels. The algorithm used to do this is called the interpolation or resampling method, and it significantly affects the quality of the result.
When you upscale an image: making it larger: the software must invent pixel values that were not in the original. Standard algorithms like bicubic interpolation do this by analysing surrounding pixels and estimating reasonable values. The result is always slightly blurry compared to a natively high-resolution image, because information that was never there cannot be recovered. AI-based upscaling tools do a better job by using trained models to predict detail, but they are not perfect.
Always maintain the aspect ratio
The aspect ratio is the proportional relationship between an image's width and height. A 1920x1080 image has a 16:9 aspect ratio. If you resize it to 800 pixels wide without adjusting the height proportionally (to 450 pixels in this case), the image stretches horizontally and subjects appear taller or squashed. Most resize tools have an option to lock the aspect ratio: always use it unless you have a specific reason to distort the image.
Choosing the right dimensions
- •Web images: aim for the largest dimension at which the image will be displayed, times two for retina screens. A 400px wide image on a webpage should be 800px wide in the source file.
- •Email: most email clients display images at a maximum of 600px wide. Resizing photos to 1200px wide (600px x2 for retina) before embedding them significantly reduces email size.
- •Social media: each platform has specific recommended dimensions. Facebook cover photos are 820x312. Instagram posts are 1080x1080. Twitter header is 1500x500. Using exact dimensions prevents unexpected cropping.
- •Print: print resolution is measured in DPI (dots per inch). A 300 DPI image for an A4 print needs to be at least 2480x3508 pixels. Resizing a small web image up to print dimensions produces a blurry result.
- •Profile pictures: circular crop platforms (Twitter, LinkedIn, Zoom) crop from the centre. Make sure the subject of the image is centred before resizing.
Resampling algorithms
Nearest-neighbour resampling is the fastest but produces the most pixelated results, acceptable only for pixel art where you want hard edges preserved. Bilinear interpolation is faster than bicubic and produces smooth results suitable for most web images. Bicubic interpolation considers a larger area of surrounding pixels and produces sharper results with better edge preservation, and is the best general-purpose algorithm for photographic content. Lanczos resampling minimises aliasing (stair-step artefacts on diagonal lines) and is preferred for line art, text, and images with fine detail.
File format after resizing
Resizing does not change the file format. If you resize a PNG, you get a PNG. If you want to change format at the same time (for example, resize a photo and convert it to WebP for web use), use a tool that handles both operations together. Doing them separately: resizing then converting: produces the same result but takes two steps.
AlteredIdea's Image Resizer lets you set width, height, or both with optional aspect ratio locking. All processing is done in your browser using the Canvas API: your image never leaves your device.