Lossless compression

Lossless compression reduces a file’s size by finding and encoding redundant patterns in the data more efficiently, without discarding any of the original information — decompressing a losslessly compressed file returns data that is bit-for-bit identical to the source. Lossy compression, by contrast, permanently removes some data as part of shrinking the file, trading a certain amount of quality for a much smaller result; that trade-off can be invisible at moderate settings but is not reversible.

In DAM practice this distinction maps directly onto the master-versus-rendition structure: the archival master benefits from being lossless because it may be re-edited, re-cropped, or re-purposed years later, and any quality loss baked into it at the master level propagates into every derivative made from it afterward. Delivery renditions, on the other hand, are usually lossy on purpose — a JPEG or standard WebP delivered to a website doesn’t need to preserve every original bit, it needs to be as small as possible while still looking correct at delivery size.

Google’s own compression research found lossless WebP images run roughly 26% smaller than equivalent PNGs at the same fidelity, which is why lossless WebP has become a practical archival alternative to PNG for graphics that need transparency or sharp edges preserved exactly. Whatever specific format a team chooses for its masters, the operating principle stays the same: the archival copy should never be the lossy one, because that decision can’t be undone later.