JPEG stands for Joint Photographic Experts Group, the committee that published the format in 1992. It is the lossy image compression standard built for photographs, and the reason almost every photo on the web (and inside social media platforms) still ends in .jpg or .jpeg. The two extensions name exactly the same format; the difference is leftover MS-DOS history from the days when Windows capped file extensions at three letters.
What does JPEG stand for?
Joint Photographic Experts Group, the international committee of standards bodies (the International Organization for Standardization, the International Electrotechnical Commission, and the International Telecommunication Union) that designed the format. The committee formed in 1986, published the JPEG standard in 1992, and is still the body that maintains the format and ships newer ones (JPEG 2000, JPEG XL, JPEG XS, JPEG XR) alongside the original 1992 release.
The standard itself is published under two names that the committee considers equivalent: ITU-T Recommendation T.81 (approved September 1992) and ISO/IEC 10918-1 (ratified in 1994). The Wikipedia entry on JPEG and the official JPEG committee's home page on jpeg.org cover the full lineage and the technical specifications.
JPG vs JPEG (the file extension question)
Same format. Two file extensions. The difference is purely historical.
Why .jpg exists at all
Early Windows (up to and including Windows 3.1) and the underlying MS-DOS used an 8.3 filename convention: up to eight characters for the name and three for the extension. The full .jpeg extension did not fit, so Windows defaulted to .jpg. macOS, Linux, and most web servers used the full .jpeg from day one because they had no such limit.
Why the convention stuck
Windows 95 lifted the 8.3 limit, but by then millions of .jpg files were already in the wild and every photo editor and digital camera defaulted to it. The shorter form became the working standard on Windows and the longer form remained the working standard on macOS. Software learned to treat the two as the same thing and never stopped.
What this means for you
Nothing practical. Any program that opens a .jpg opens a .jpeg without converting anything; every social platform accepts both extensions on upload and stores them as the same file. The choice between the two is cosmetic, and most teams pick one for consistency in filenames rather than for any technical reason.
How JPEG compression actually works
The format is built around three working ideas, all of them designed to throw away data the human eye is least likely to notice while keeping the data it is most sensitive to.
Discrete cosine transform (DCT)
The maths underneath JPEG. The image is divided into 8x8 pixel blocks, each block is transformed from spatial colour values into a frequency representation, and the high-frequency components (the fine detail) are aggressively quantised and discarded. The eye is much better at picking out big shapes and smooth gradients than tiny detail, so throwing away the high frequencies first is the right trade for visual fidelity.
Chroma subsampling
Human eyes are more sensitive to brightness than to colour, by a wide margin. JPEG exploits this by storing the brightness channel at full resolution and downsampling the colour channels by half (the 4:2:0 ratio used in most JPEGs throws away 75% of the colour information). The result is a file that looks almost identical to the eye while weighing a fraction of the original.
The quality slider
The single most important setting when saving a JPEG. Quality 100 is near-lossless and produces files almost as large as the original. Quality 90 to 95 is the sweet spot for source files that may be re-compressed by a platform. Quality 75 to 85 is the working band for direct web use, where files are 5 to 10 times smaller than the original with a visible difference only a careful viewer would notice. Quality 60 and below produces visible 8x8 blocking and colour banding.
Compression ratio
JPEG typically achieves around 10:1 compression with a perceptible but acceptable quality loss, which is the figure on the official standard and the rough ceiling for everyday use. Photos in particular compress well; text, line art, and screenshots compress poorly because the format is built for smooth-gradient content, not for hard-edged graphics.
A short history of JPEG
1986: The committee forms
The Joint Photographic Experts Group is established as a working group inside the ISO and CCITT (the predecessor of ITU-T) to design a single international standard for compressing colour and greyscale photographs. The committee runs through five years of competing proposals before settling on the DCT-based design that becomes JPEG.
1992: The standard ships
ITU-T approves Recommendation T.81 in September 1992. ISO/IEC ratifies the equivalent standard as ISO/IEC 10918-1 in 1994. The format reaches the consumer web almost immediately through Mosaic (the first widely-used graphical browser) and becomes the dominant photographic format on the early web alongside GIF.
Mid 1990s to 2000s: Digital photography
JPEG arrives at exactly the right moment for digital photography. The first consumer digital cameras ship with JPEG as the primary format, the first phone cameras follow, and by the early 2000s almost every photo on a hard drive or a website is a JPEG. The format weathers the transition from CRT to LCD to high-density display almost unchanged.
2010s: HEIC arrives on iPhone, WebP arrives on the web
Apple ships HEIC (High Efficiency Image Container) as the default photo format on iPhone in iOS 11 in 2017, with file sizes around half the equivalent JPEG. Google ships WebP in 2010 and grows its browser support through the 2010s to near-universal. Both formats deliver better compression than JPEG; both struggle to dethrone it because every existing camera, social platform, and CMS already speaks JPEG natively.
2020s: AVIF and JPEG XL
AVIF (the still-image variant of the AV1 video codec) and JPEG XL (the JPEG committee's modern replacement for the original format) both arrive in the early 2020s and both deliver dramatic compression and quality improvements over the 1992 standard. JPEG XL is backwards-compatible with classic JPEG (existing files can be converted losslessly), but adoption is slow because the install base of JPEG is one of the largest in computing history.
JPEG vs PNG, WebP, AVIF, and HEIC
The working choice between formats in 2026.
JPEG: photographs
Lossy, no transparency, 8-bit colour, universally supported. The right format for photographs, photo-style art, complex gradients, and anything where smooth colour shifts matter more than hard edges. The wrong format for logos, screenshots, line art, anything with text on a flat background, and anything that needs alpha transparency.
PNG: screenshots, logos, transparency
Lossless, supports a full alpha channel, 8-bit or 16-bit colour, universally supported. The right format for screenshots, logos, icons, line art, and any image with text on a solid background. A photo saved as PNG is almost always 5 to 10 times the file size of the equivalent JPEG with no visible quality gain.
WebP: the modern default
Google's modern format, around 25 to 35% smaller than JPEG at the same quality, with support for both lossy and lossless modes, alpha transparency, and animation. Over 95% browser support globally in 2026, which puts it past the threshold where a fallback to JPEG is necessary for nearly any audience. The right default for new web work.
AVIF: the smallest files
An even newer format based on the AV1 video codec, around 50% smaller than JPEG (and 20 to 30% smaller than WebP) at the same quality. Slightly narrower browser support than WebP in 2026 (around 94%) but rising fast. The right format for performance-critical work where every kilobyte matters; use it with a WebP or JPEG fallback for the long tail of clients.
HEIC: iPhone's default
Apple's container format, based on the HEVC video codec, around half the file size of JPEG at the same quality. The default photo format on iPhone from iOS 11 onwards. Universally supported across Apple devices and modern macOS; less well supported on Windows and Android, which is why HEIC photos often get auto-converted to JPEG on upload to most social platforms.
JPEG XL: the long-term replacement
The JPEG committee's own modern format, designed to be losslessly convertible from classic JPEG (existing files can be re-encoded without any further quality loss). Around 60% smaller than JPEG at the same quality, with full alpha support, HDR, and progressive decoding. Browser support is the bottleneck in 2026: Safari ships it, Firefox supports it behind a flag, Chrome removed support in 2022 and has not added it back. Adoption is real but slow.
JPEG on social media in 2026
Almost every social platform accepts JPEG (and PNG, and WebP on most, and HEIC on some), and almost every social platform re-compresses every upload. The practical consequences for anyone uploading images:
- Upload at higher quality than the final spec. The platform will re-compress the file on upload, which runs the lossy compression a second time on top of the quality setting you exported at. Saving the source at quality 90 to 95 means the platform's re-compression lands on a still-good file; saving at quality 75 means the platform's re-compression lands on artefacts on top of artefacts.
- Save the right dimensions for each platform. Uploading a 4000 px photo to Instagram, which displays it at 1080 px in the feed and 1440 px in stories, is a waste of bytes and trips the platform's re-compression unnecessarily. The right move is to resize to the platform's display size (or to its safe upper bound) before export.
- Pick the right format for the right content. Photos go as JPEG. Logos, screenshots, infographics with flat colour and text go as PNG. The wrong format produces either bloated files (a screenshot as JPEG with visible artefacts around the text, then re-compressed by the platform) or huge files (a photo as PNG, which the platform will silently re-encode to JPEG anyway, with the double-conversion making it look worse than the source).
- Avoid edge-case quirks. A few platforms have quirks worth knowing: Pinterest handles very long Pin images (up to 1000 x 1500 px) well but strips most metadata; X re-compresses uploads aggressively and looks better on PNG for graphics and JPEG for photos; LinkedIn keeps fairly high quality on first upload but degrades on reshares; Threads inherits most of Instagram's compression rules. The platform spec sheets change often enough that a yearly check is worth scheduling.
- Keep an uncompressed master. A PSD, TIFF, RAW, or full-quality PNG export of the source image, kept somewhere safe, means a fresh JPEG can be exported every time. Editing a JPEG, re-saving it, and uploading the re-saved file is the surest path to compounding artefacts across versions.
Common JPEG mistakes
- Re-saving a JPEG over itself. Every save runs the lossy compression again, and the artefacts compound. The fix is to keep the source file (PSD, TIFF, RAW, or PNG) and export a fresh JPEG each time, rather than opening a JPEG, editing it, and saving it back over the original.
- Saving logos and screenshots as JPEG. The JPEG compression is built for smooth-gradient photos, not for the hard edges and flat colour fields of logos and screenshots. The result is visible colour fringing around text, blocky artefacts on the edges of flat shapes, and a file that is often larger than the equivalent PNG would have been. Save these as PNG.
- Forgetting that JPEG does not support transparency. A transparent area in the source becomes a flat-colour background in the JPEG, almost always white. Anything that needs to sit on a coloured background without a visible rectangle around it needs to be a PNG or a WebP, not a JPEG.
- Cranking the quality slider to 100 by default. Quality 100 produces files near the size of the original with almost no visual benefit over quality 85 to 90. The right setting for source files is in the 90 to 95 range; the right setting for direct-to-web JPEGs is in the 75 to 85 range. Quality 100 is the answer to a problem most uploads do not have.
- Uploading a HEIC photo without converting it. A HEIC file from an iPhone often fails to upload to surfaces that do not handle the format (older CMS, some email clients, the occasional embed). The fix is to export the photo as JPEG (or WebP) before uploading, rather than letting the destination platform's on-the-fly conversion decide the quality settings.
- Stripping the wrong metadata. A JPEG carries EXIF metadata (camera, lens, exposure, GPS, date) and ICC colour-profile metadata. Stripping the EXIF for a public upload is usually a privacy win (GPS coordinates in particular). Stripping the ICC colour-profile metadata is usually a colour-fidelity loss, because browsers and platforms render the image against an assumed sRGB profile that may not match the photo. The cleanup pass should strip EXIF and keep ICC.
- Resizing in the wrong order. Compress, then resize is the wrong order: the compressed file gets resized with the compression artefacts already baked in, and the second pass makes them worse. Resize first (to the final display dimensions or a sensible upper bound), then compress, then export. Photo software almost always does this correctly when given the right export preset; manual edits are where the order goes wrong.
For the glossary entries this one connects to, the GIF entry covers the close cousin built for animation rather than photos, the emoji entry covers the small-vector cousin most chat surfaces layer on top of JPEG-and-PNG content, the watermark entry covers the attribution layer most JPEGs ship with on social media, and the carousel post entry covers the multi-JPEG format Instagram and LinkedIn run on.
The matching tools on this site cover the working adjacent work. The social image resizer handles the dimensions for every major platform in one upload, the alt-text generator writes the accessible caption a screen reader will read for the JPEG, and the Instagram grid planner lines the JPEGs up on the profile before they go live.
JPEG FAQ
What does JPEG stand for?
JPEG stands for Joint Photographic Experts Group, the committee of standards bodies (ISO, IEC, and the ITU) that created the format. The committee formed in 1986 and published the JPEG standard in 1992, which was then ratified as ITU-T Recommendation T.81 in September 1992 and as ISO/IEC 10918-1 in 1994. The committee still maintains the standard and ships newer formats (JPEG XL, JPEG XS, JPEG XR) alongside the original.
Is JPG the same as JPEG?
Yes. JPG and JPEG are the same file format with two different file extensions. The .jpg shortening goes back to the MS-DOS and early Windows 8.3-character filename limit, which capped file extensions at three letters; macOS, Linux, and most web servers shipped .jpeg from the start. Modern operating systems do not care about the difference, and any program that opens a .jpg will open a .jpeg without converting anything. The two extensions are interchangeable in 2026.
Is JPEG lossy or lossless?
Lossy, in almost every common use. The format throws away data that the human eye is least likely to notice (high-frequency detail and colour information), which is why a 5 MB photo compresses to a 500 KB JPEG. The format does have a lossless mode in the original 1992 standard, but it is almost never used because the file-size advantage is small compared to dedicated lossless formats like PNG. The trade most people sign up for is a small visible quality loss in exchange for a tenfold reduction in file size.
What quality should I save JPEGs at?
For the web and social media, JPEG quality 75 to 85 in Photoshop or any modern tool is the working band: visually almost indistinguishable from the original for most photos, with file sizes 5 to 10 times smaller than the source. Quality 90 to 95 is the right setting for content that may be re-uploaded or re-compressed by the platform (most social platforms re-compress every upload), because it survives one or two more passes before the artefacts become visible. Quality 60 and below is the band where the blocky 8x8 compression artefacts and the colour banding become obvious to a casual viewer.
JPEG vs PNG, which is better?
Neither, separately. JPEG is the right format for photographs and complex images with smooth colour gradients, because the DCT compression is built for that kind of content and file sizes stay small. PNG is the right format for screenshots, logos, line art, icons, and any image that needs transparency, because the compression is lossless and PNG supports an alpha channel. A photo saved as PNG is almost always a much larger file with no visible quality gain over JPEG; a logo saved as JPEG always has visible artefacts and colour fringing around the edges.
Should I use JPEG or WebP on my website?
WebP, almost always, with a JPEG or PNG fallback for the small share of clients that still cannot read it. WebP delivers roughly 25 to 35% smaller files than JPEG at the same visual quality, supports both lossy and lossless modes, supports an alpha channel, and ships with browser support over 95% globally in 2026. The decision tree is: use AVIF for the smallest files (around 50% smaller than JPEG) with a WebP fallback, use WebP as the default with a JPEG fallback, use JPEG only when the surface (an old chat client, an email signature, a legacy CMS) cannot accept WebP.
Why does my JPEG look worse every time I save it?
Because every save runs the lossy compression again, and the artefacts compound. The first save throws away the data the format's compression algorithm decides is least visible; the second save runs the same process on the already-compressed file, which makes the algorithm decide a different set of bits is least visible, and so on. The fix is to keep an uncompressed master file (PSD, TIFF, RAW, or even a high-quality lossless PNG), edit only the master, and export a fresh JPEG every time, rather than opening a JPEG, editing it, and re-saving it back over itself.