Treating an exported JPEG as the master and editing it repeatedly, which stacks a new round of quantisation damage on every save.
Compression
Lossy compression
Also called lossy or lossy image compression. Here is what it means, when it changes what you export, and what people get wrong.
CompressionRuns on your deviceUpdated 10 August 2026
The short answer
Lossy compression shrinks an image by permanently throwing away detail your eye is least likely to notice. The discarded data cannot be recovered, so every save is a one-way trip. JPEG, and the lossy modes of WebP and AVIF, all work this way, trading measurable accuracy for much smaller files.
Updated
The short version
Lossy compression at a glance
| Quick fact | Detail |
|---|---|
| Reversible | No, discarded detail is gone permanently |
| Controlled by | A quality or quantisation setting |
| Formats | JPEG, WebP lossy, AVIF, HEIC |
| Best on | Photographs and natural texture |
| Worst on | Flat colour, text, logos, screenshots |
Slide the table sideways to see every column.
How does a lossy encoder decide what to discard?
JPEG, the format most people mean when they say lossy, works in three moves. It splits the picture into 8x8 pixel blocks and applies a discrete cosine transform, which rewrites each block as a set of frequency coefficients instead of raw pixel values. Low frequencies describe the broad shape of the block, high frequencies describe fine texture and sharp edges.
It then divides those coefficients by a quantisation table. Large divisors are used on the high-frequency coefficients, so most of them round to zero. That rounding is the loss, and it is irreversible. What remains is a much shorter list of numbers, which is packed with entropy coding (Huffman or arithmetic) into the final file.
Most lossy encoders also apply chroma subsampling, storing colour at half resolution because human vision is far more sensitive to changes in brightness than in colour. Newer lossy codecs such as WebP and AVIF use larger, smarter block structures and better prediction, but the underlying bargain is identical.
Lossy or lossless: which one fits the job?
| Property | Lossy | Lossless |
|---|---|---|
| Pixels after decoding | Approximate | Identical to the original |
| Typical size on a photograph | Much smaller | Only modestly smaller |
| Repeated resaving | Degrades each time | Safe indefinitely |
| Good for | Photographs, natural texture, gradients | Logos, screenshots, text, line art, masters |
| Common formats | JPEG, WebP lossy, AVIF, HEIC | PNG, WebP lossless, TIFF, lossless AVIF |
The dividing line is content, not quality standards. A photograph has so much fine random variation that lossless coding has little to exploit, and lossy coding hides its errors in that same variation. A flat white product background with a sharp logo on it is the reverse case, and lossy coding puts its errors exactly where you will see them.
Why does re-saving a JPEG keep making it worse?
This is generation loss. Each time you open a lossy file and save it again, the encoder quantises data that has already been quantised once. If the image has been cropped, rotated by an arbitrary angle, resized or edited, the 8x8 block grid no longer lines up with the previous pass, so a fresh set of rounding errors lands on top of the old ones.
The practical rule is to keep an untouched master in a lossless format, do all your editing on that, and export a lossy copy only as the final step for each destination. If you need three sizes for a listing, export three times from the master rather than resaving the export.
What people get wrong about lossy compression
Each one is a real failure mode, not a style preference.
Assuming quality 100 means lossless, when it only means the mildest quantisation the encoder offers.
Using lossy compression on a logo or a screenshot, where ringing and colour bleed around hard edges are immediately obvious.
Judging compression damage on a zoomed-out preview rather than at the size the image will actually be viewed.
Questions people ask
Lossy compression, answered
The follow-up questions people search for once they have the definition.
Does opening a JPEG and closing it without saving damage it?
No. The loss happens when the encoder writes the file, not when a viewer reads it. You can open and view a JPEG as many times as you like. Damage only accumulates when you save a new copy through a lossy encoder.
Can you undo lossy compression?
No. The quantisation step rounds real values to zero and there is no record of what was there before. Denoising and AI upscaling can make artefacts less obvious by inventing plausible replacement detail, but they are not recovering the original pixels.
Is lossy compression bad for print?
Not automatically. A high quality lossy export at full print resolution prints fine for most photographic work. The risk is compounded loss from repeated edits, and artefacts around type or logos, which show up harshly on paper at 300 PPI.
How much smaller is lossy than lossless?
It depends entirely on the picture. On a detailed photograph, lossy files are usually several times smaller than a lossless version at a quality most people cannot distinguish. On flat graphics the gap narrows sharply, and lossless can even win.
Same cluster
More from the glossary
Neighbouring entries, so every term in the set is one hop from every other.
Mask
A mask is a non-destructive control that decides which parts of a layer are visible without deleting any pixels.
Matte
A matte is a greyscale image that records how opaque every pixel of a subject is: white for fully solid, black for fully transparent and grey for everything in between.
Listing image
A listing image is a photo attached to a product listing on a marketplace.
From the studio
Tools that do this
Each one runs on your own device, so the photo stays where it is.
Sources
Where the facts on this page come from. Every link opens the specification, standard or documentation the claim was read at.
Keep reading
Elsewhere on Photomop
Platform image rules
Image size, file type and background rules for 22 selling platforms, each one sourced.
Read itConvert image
JPG, PNG, WebP and AVIF, with an honest answer about HEIC and alpha.
Read itAll the tools
The whole toolbox in three families, with what each one needs before it can run.
Read itOn-device privacy
The pipeline in five steps, what leaves your device, and how to check it yourself.
Read it
Now do it to a photo
Photomop is a photo studio that runs on your own device. Resize, crop, compress, convert, batch edit, remove a background and change one all work in the browser tab you are reading this in, at full resolution and with no watermark. The on-device page shows you how to check that the photo stays put.