Image Resize Calculator

Proportional resizing keeps the aspect ratio, width divided by height, constant. Given the original dimensions and a target width, the new height is the original height multiplied by the ratio of new width to original width. Reducing 1920 by 1080 to a width of 1280 gives 1080 times 1280/1920, or 720. Expressing the ratio in lowest terms by dividing both dimensions by their greatest common divisor yields the familiar names: 1920 by 1080 reduces to 16:9, 1024 by 768 to 4:3, and 1080 by 1350 to 4:5. Scaling is a resampling operation rather than a simple selection of pixels, so the filter matters. Nearest neighbor is fast but produces blocky edges, bilinear interpolation averages four neighboring samples, bicubic uses sixteen and holds edges better, and Lanczos windowed sinc filtering gives the sharpest downscales at the cost of ringing near high-contrast borders. Downscaling without first applying a low-pass filter causes aliasing, visible as moire patterns on fine repeated detail such as fabric weave or brickwork. The asymmetry between directions is the central limitation: downscaling discards information and generally looks clean, while upscaling can only interpolate detail that was never captured, so enlarging beyond the source resolution softens the image no matter which filter is chosen. Common targets are a maximum of about 1920 px wide for full-bleed web imagery, around 600 px for email templates, and for print the intended physical size in inches multiplied by 300 pixels per inch.

Aspect ratio conventions were set by film and broadcast long before digital imaging. The 4:3 frame descends from the 35 mm format developed by Thomas Edison and William Kennedy Dickson in the early 1890s, was adopted as the Academy ratio in 1932, and carried into NTSC and PAL television. Widescreen cinema formats proliferated in the 1950s, and in the 1980s Kerns Powers of the David Sarnoff Research Center derived 16:9 as a compromise that best accommodated the range of formats then in use. SMPTE and later the ATSC digital television standard adopted it, and computer displays followed in the mid-2000s. The mathematics of resampling came from signal processing. The sampling theorem associated with Harry Nyquist work in 1928 and Claude Shannon 1949 paper explains why an image must be band-limited before reduction, or high-frequency detail folds back as aliasing. Bicubic convolution interpolation was formalized by Robert Keys in a 1981 IEEE paper, and the Lanczos filter applies a windowed sinc function named after the mathematician Cornelius Lanczos. These techniques reached ordinary users through desktop software. Adobe Photoshop 1.0 shipped in 1990 with resampling built in, and the open-source ImageMagick, first released by John Cristy the same year, made scriptable batch resizing widely available. Browser-side responsive images came much later, with the srcset and sizes attributes standardized around 2014.