Lossy

Lossy refers to a type of data compression in which some information from the original data is discarded to reduce file size. In lossy compression, the exact original data cannot be reconstructed; the result is an approximation that aims to preserve perceptual quality rather than exact fidelity. Lossy methods are widely used for multimedia to enable efficient storage and transmission.

Most lossy compression relies on perceptual models that exploit limits of human perception. The typical workflow transforms data into a domain such as frequency or time-frequency space, applies quantization to reduce precision of coefficients, and then uses entropy coding to compress the result. The degree of quantization controls the amount of information removed, creating a trade-off between bit rate and reconstructed quality. Common lossy formats include JPEG for images, MP3 and AAC for audio, and H.264/AV1 and MPEG-4 for video.

The advantages of lossy compression are significantly smaller file sizes and lower bandwidth requirements, enabling streaming and convenient storage of large media collections. The main drawbacks are artifacts that can appear when data is aggressively compressed, such as blocking, blurring, ringing, and color banding. Artifacts depend on content, bitrate, and the codec used, and they may be minimized by higher bitrates or improved perceptual models.

Quality assessment of lossy compression often combines objective metrics such as PSNR or SSIM with perceptual testing, since human observers primarily judge visual and auditory quality. In practice, rate-distortion optimization guides how codecs allocate bits to different parts of the signal to balance size and fidelity. Lossy compression is contrasted with lossless compression, which preserves exact original data but yields larger files.