Reference Glossary
Lossy compression
A compression method that permanently discards some image, audio, or video data to shrink file size, trading fidelity for smaller storage and faster delivery.
Why it matters in a DAM
DAM platforms rely on lossy formats like JPEG for preview and delivery renditions because they cut storage and bandwidth costs dramatically across large libraries. The catch is that discarded data cannot be recovered later — re-saving an already-compressed lossy file compounds visible artifacts each time it's edited and re-exported, which is why a DAM's master should never itself be a lossy derivative.
A worked example
Common mistake
Uploading a JPEG as the master asset instead of the original TIFF, RAW, or PSD, so every derivative generated afterward inherits and compounds an already-lossy source.
Lossy compression works by identifying data a viewer is least likely to notice missing and discarding it outright. JPEG, the most common lossy image format, does this by averaging colour detail across blocks of pixels (chroma subsampling) and simplifying fine texture — the more aggressive the setting, the more data is thrown away and the smaller the file. Once discarded, that data is gone; opening a heavily compressed JPEG and re-saving it as a new master doesn’t restore anything, it just adds a fresh round of loss on top of the first.
In a DAM, this makes the distinction between master and derivative operationally important, not just academic. A master asset should be lossless or a RAW/camera-original file, stored once and never re-compressed. Everything a user actually downloads — a web preview, a social crop, a print proof — should be generated fresh from that master through a lossy pipeline tuned to the use case. If the master itself is lossy, every derivative made from it inherits that ceiling on quality, and repeated internal edit-and-resave cycles (common when assets get routed through multiple creative tools before final approval) visibly degrade the image over time, an effect known as generational loss.
Frequently asked
What is lossy compression?
Lossy compression is a method that permanently discards some image, audio, or video data in order to shrink file size, trading fidelity for smaller storage and faster delivery. This is fundamentally different from lossless compression, which shrinks a file by encoding its data more efficiently without removing any of it, allowing the original to be reconstructed exactly. Once lossy compression discards data, that information is gone for good and cannot be recovered by any later process.
How does JPEG achieve lossy compression?
By averaging colour detail across blocks of pixels (chroma subsampling) and simplifying fine texture — the more aggressive the setting, the more data is thrown away and the smaller the file.
Why should a DAM master never be a lossy file?
Discarded data cannot be recovered later, so if the master itself is a lossy file, every derivative generated from it inherits that same ceiling on quality with no way to work back to a sharper version. Each time that lossy master is re-saved or re-compressed during editing, it also accumulates fresh generational loss on top of what was already discarded. A DAM master should instead be the single lossless source that every derivative is generated from, preserving maximum quality for any future use case.
What is generational loss?
The visible degradation that accumulates when a lossy file is repeatedly opened and re-saved, common when assets get routed through multiple creative tools before final approval, since each re-save adds a fresh round of loss on top of the last.
What's the common mistake with master uploads?
The common mistake is uploading an already-compressed JPEG as the master asset instead of retaining the original TIFF, RAW, or PSD file. Once only the JPEG exists in the system, the lossless source is gone for good — there is no way to reconstruct the discarded data later, no matter how the file is processed. Every derivative generated afterward then inherits and compounds that already-lossy source, permanently capping the quality ceiling for all future uses of the asset.
What should generate lossy delivery files in a DAM workflow?
A rendering pipeline should generate lossy delivery files on the fly, producing a lossy version tuned to the specific use case fresh from the lossless master each and every time one is needed. The master itself should never be that permanently lossy file; it stays lossless so it can keep serving as the single accurate source for every future derivative, rather than the workflow saving and reusing an already-compressed output as if it were the original master asset.