Lossy compression discards data we cannot perceive to make digital files significantly smaller.

Science · Published 20 October 2026 · Updated 21 October 2026 Swipe more like this →

In short: Digital files often contain more information than human senses can process. Lossy compression exploits this by stripping away redundant details, allowing for massive reductions in file size. While this process is irreversible and can degrade quality, it powers the streaming services and internet communication that define our modern digital experience.

At its core, lossy compression is an exercise in selective discarding. By utilizing inexact approximations, it reduces the bandwidth and storage required for multimedia content like audio, video, and images. Unlike lossless methods, which preserve every bit of original data, lossy techniques rely on the fact that human perception is limited. For instance, the human eye has varying sensitivity to color components, and the human ear does not require fine detail during loud audio passages. By matching compression to these physiological idiosyncrasies, engineers can remove information that the end-user will likely never notice.

The most prevalent method for this is transform coding, specifically the discrete cosine transform (DCT), first published in 1974 by Nasir Ahmed, T. Natarajan, and K. R. Rao. This technique transforms data into a domain—such as the frequency spectrum for audio—that better aligns with how we perceive information. Once transformed, the data is quantized, a process that selectively removes the least significant information. This approach is highly effective; video can often be compressed at ratios as high as 100:1 with little visible quality loss, while audio and still images typically see 10:1 ratios.

Despite its efficiency, lossy compression carries the risk of generation loss. Repeatedly compressing and decompressing a file progressively degrades its quality, a phenomenon that can be avoided by editing only from original, lossless masters. While some minor operations like cropping or rotating JPEGs can be performed losslessly without re-encoding, most edits require caution. Ultimately, lossy compression is a calculated tradeoff, governed by rate-distortion theory, which uses Bayesian estimation to model perceptual distortion. It turns the digital world into a manageable stream, prioritizing utility and speed over the preservation of every original bit.

General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.

Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?

Related

• Why your digital music files are not just raw audio data
• How computers shrink data without losing its essence
• How do we shrink massive amounts of data into smaller files?
• How do we shrink massive audio files for seamless streaming without losing quality?
• How your devices shrink massive amounts of data into tiny files
• The London stage that first conquered the darkness with gas
• The universe may have started with a bounce, not a bang
• How a refugee shield-maker became the architect of the world's most influential plain prose.
• How a nineteenth-century railway company reshaped London's geography and pioneered the commuter age
• Can we unify gravity with the quantum world of the very small?
• Scotland's local government is a complex tapestry of thirty-two unitary authorities and ancient roots.
• Mac OS 8 was a commercial triumph born from the ashes of a failed project.

More in Physics · More in Science · Latest · A–Z · All topics