JPG files: what the compression actually does

How JPEG discards detail, why re-saving degrades an image, and when PNG or WebP is the better choice. Convert JPG in your browser, no upload.

At a glance

Type: Image
Compression: Lossy
Extensions: .jpg, .jpeg
MIME type: image/jpeg
Also known as: JPEG

JPEG, standardised in 1992, divides an image into 8×8 blocks and transforms each one into frequencies rather than pixels. High frequencies — abrupt changes between neighbouring pixels — are quantised away first, because human vision is least sensitive to them. That single design decision explains everything about the format: it is superb on photographs, where detail varies smoothly, and visibly bad on text and line art, where the abrupt changes it discards are the entire content.

It also explains generation loss. Each save re-quantises an image that has already been quantised, so artefacts compound: the faint ringing around hard edges gets stronger, and flat areas develop the blocky mottling that gives away a much-re-saved image. Editing a JPG and saving it repeatedly is the most common way people quietly destroy their own photographs. Convert to PNG first if you are going to work on an image over several sessions.

JPEG has no alpha channel, so transparency is not merely unsupported — there is nowhere to put it. Any conversion into JPG has to decide what transparent pixels become, and this site fills them with white, which is right for a photograph on a light page and wrong for a logo. It is also 8 bits per channel in normal use, which is why gradients in a JPG sometimes band where a 16-bit source did not.

On naming: .jpg and .jpeg are the same format, the three-letter version surviving from MS-DOS filename limits. What actually varies between JPG files is the metadata wrapper — JFIF, EXIF, or both — which carries camera settings, orientation and often GPS coordinates. Converting through a browser canvas, as this site does for image work, discards all of it: a privacy benefit, though it means the orientation tag is applied to the pixels rather than preserved.

Convert JPG to other formats

Convert other formats to JPG

FAQ

What quality setting should I use?

82 is the default and a good balance for photographs — visually close to the original at a fraction of the size. Screenshots and images containing text suffer more, so use 90 or above for those, or keep them as PNG.

Does re-saving a JPG make it worse?

Yes, every time. Each save re-quantises already-quantised data, so ringing around edges strengthens and flat areas become blotchy. Always re-encode from the original rather than from a previous JPG.

JPG, PNG or WebP — which should I choose?

JPG for photographs where compatibility matters most. PNG for text, screenshots, diagrams and anything needing transparency. WebP for the web, where it is 25–30% smaller than JPG at the same visible quality and supports transparency too.

Why does my JPG look blocky?

Because the quality setting was low, or because it has been saved repeatedly. JPEG works in 8×8 blocks, and when there are too few bits to describe them the block boundaries become visible.

What happened to my transparent background?

It became white. JPEG has no alpha channel at all, so transparency cannot be represented. Convert to PNG or WebP if you need to keep it.

Is .jpg different from .jpeg?

No, they are the same format. The three-letter extension is a leftover from MS-DOS filename limits. Both are accepted here.

Is EXIF and GPS data kept?

No. Image conversions here run through the browser's canvas, which does not carry metadata across, so camera details and location are stripped. Orientation is applied to the pixels rather than preserved as a tag.

Does converting a JPG here upload it?

No. Your browser already contains JPEG codecs, so the work happens in the page with nothing to download and nothing sent. It is the fastest kind of conversion on this site.