Image glossary
This glossary explains the technical vocabulary of digital images in plain language. Whether you are compressing a photo, converting a format, or stripping metadata, these terms give you the foundation to understand what the tools are doing. Each entry takes under a minute to read, with no heavier math than necessary and no jargon left unexplained.
Definitions favour the way terms are used in web workflows, so you get practical context rather than a computer-science primer. Where a term has a close cousin, such as DPI versus PPI, the entry steers you to the related definition so you can compare them. Jump to any letter below or browse the whole list whenever you need a refresher.
Alphabetical index
A
Alpha
The alpha channel stores opacity for every pixel, from fully transparent to fully opaque. Photographs rarely use alpha, while logos, icons, and interface graphics depend on it to blend over any background. Without it, an image always shows as a solid rectangle.
Animated GIF
A GIF holding multiple frames that a browser plays in sequence. Because it is limited to 256 colors, animated GIFs suit small flat-colored clips but bloat on photographic content. WebP and AVIF animate just as easily with much smaller files.
Aspect ratio
The proportional width-to-height relationship of an image, written with a colon such as 16:9. Changing it requires cropping or padding, never stretching, because distortion destroys the natural look of a photo. Framing and cropping choices are governed by this ratio.
AVIF
A modern format built on the AV1 video codec with outstanding compression, often beating WebP by 20 to 30 percent at equal quality. It supports HDR, lossy and lossless modes, transparency, and animation. Browser support now covers most users, making it an excellent web choice with a WebP or JPEG fallback.
B
Base64 / Data URI
Base64 is a text encoding that writes binary data with a safe 64-character alphabet. A data URI is the resulting payload embedded directly into HTML, CSS, or JSON. It saves a network request but adds about 33 percent overhead, so reserve it for tiny icons and sprites.
Bicubic interpolation
A resampling algorithm that weighs the sixteen nearest source pixels when resizing. It keeps tonal transitions smooth, making it the standard for photographs. Sharper variants emphasize edges, while the plain version is a balanced default.
Bilinear interpolation
A faster resampling algorithm that blends the four nearest pixels. It handles shrinking well but softens fine detail when enlarging. Good for thumbnails and previews where speed beats perfect sharpness.
Bit depth
The number of bits holding color per pixel, which fixes how many tones an image can show. An 8-bit channel gives 256 levels per color, over 16 million colors total, the web baseline. Deeper 16-bit channels smooth gradients and survive editing better.
BMP
The uncompressed Windows Bitmap format from the early PC era. Every pixel is stored with no compression, so files are enormous and useless on the web. Converting BMP to PNG or JPG usually shrinks the file dramatically.
C
Chroma subsampling
A compression technique that stores color at lower resolution than brightness. Because the eye sees luminance better than color, dropping color detail saves space with little visible cost in most photos. Labels like 4:4:4 and 4:2:2 describe how much color survives.
Color gamut
The full range of colors a device can show or a format can record. sRGB covers a modest gamut tuned for the web, while Adobe RGB and Display P3 hold more vivid colors. Moving files between gamuts can shift their appearance.
Color profile
A standardized description of how color values should be interpreted, stored inside the file as an ICC profile. Cameras, monitors, and printers each read numbers differently, and the profile keeps colors consistent across them. Converting to sRGB prevents cross-device color surprises on the web.
Color space
The mathematical model mapping color values to visible hues, such as RGB, CMYK, or HSL. Each space describes color in different coordinates, so the same numbers render differently in each. Pick the space to match the output, sRGB for web and CMYK for print.
Compression
Reducing the bytes an image needs by encoding its data more efficiently. Lossless compression removes redundancy without losing pixels, while lossy compression discards detail viewers rarely notice. Good tools balance the two around the intended viewing size.
Contrast
The brightness difference between the light and dark areas of an image. High contrast looks punchy; low contrast looks flat. It also drives accessibility, since text and background must hold a ratio near 4.5:1 to stay readable.
Cropping
Removing the outer edges of an image to reframe the subject or change its shape. Cropping deletes pixels permanently, so do it near the end of your workflow. It is the correct way to change aspect ratio, unlike stretching, which distorts content.
D
DCT
The Discrete Cosine Transform, the math core of JPEG that converts a block of pixels into frequency components. Low frequencies carry broad tonal structure; high frequencies carry fine detail that can be reduced. This is how JPEG removes invisible detail rather than random pixels.
Dithering
A technique that fakes missing colors by arranging available ones in a careful pattern. It helps when reducing to small palettes, like GIF’s 256 colors, and when hiding gradient banding. It trades a little fine detail for a smoother overall look.
Downscaling
Reducing an image’s pixel dimensions to a smaller size, trimming bytes and processing cost. It is generally quality-safe because removed detail is often invisible at the smaller size. It is the most effective single optimization for web images.
DPI
Dots per inch, a print measurement of how densely a printer lays ink. It is often confused with PPI, the pixel density of an image file, and neither affects on-screen display. The DPI tag is only a sizing hint for printing.
E
EXIF
Exchangeable Image File Format, the metadata standard cameras embed in photos. It records exposure settings, dates, device model, and often GPS coordinates of the shot. That location data is a privacy risk when sharing, so stripping EXIF is a core privacy habit.
F
File signature
The fixed bytes at the start of a file identifying its format, also called magic bytes. Browsers and systems trust the signature over the filename extension when deciding how to open a file. This is how corrupted or mislabeled files get diagnosed.
Flat colors
Uniform areas of a single color with no gradient, typical of logos and diagrams. They compress extremely well in PNG and lossless WebP. Compressing them with a photographic format can create visible artifacts even at high quality.
G
Gamma
The brightness curve applied so the limited tones of a display feel evenly spaced to human eyes, which sense light non-linearly. Without gamma encoding, gradients look harsh and shadows crush. Matching gamma to your target medium prevents dark or washed-out images.
GIF
Graphics Interchange Format, a 1987 bitmap format limited to 256 colors per frame with lossless LZW compression. It supports simple animation and became the classic looping-clip format. WebP and AVIF deliver the same animation in far smaller files.
Grayscale
An image mode holding only brightness, so every pixel is a shade of gray. Files shrink because each pixel stores one channel instead of three. It works as a style choice, a size saver, and the basis of contrast analysis.
H
HEIC
The default format modern iPhones use for photos, built on the HEVC video standard. It runs about half the file size of an equivalent JPEG and supports Live Photo frames. Support outside Apple devices is limited, so sharing usually means converting to JPG or WebP.
I
ICC
International Color Consortium, the body behind the standard color profile embedded in image files. An ICC profile maps device-specific color numbers into a device-independent space. With it, the same image looks consistent across cameras, monitors, and printers.
Interpolation
Calculating new pixel values when resizing, rotating, or transforming an image. Algorithms estimate missing pixels by blending neighbors, trading speed against quality. Simple ones blur; advanced methods like bicubic preserve edges at greater cost.
IPTC
International Press Telecommunications Council metadata for attaching editorial information to photos. Its fields cover captions, keywords, credits, copyright, and location. Newsrooms and stock agencies rely on it for licensing and retrieval.
J
JPEG
The most widespread photographic format, named for the Joint Photographic Experts Group. Its lossy algorithm discards subtle color and detail, reaching 10 to 20 times the size reduction with little visible loss. It supports neither transparency nor animation, so it suits continuous-tone photos.
L
Lossless
A compression mode removing redundancy while reconstructing every original pixel exactly. It is ideal for archival, text-heavy graphics, and zero-loss workflows. The trade-off is weaker ratios, typically 50 to 80 percent savings rather than 90 percent or more.
Lossy
A compression mode permanently discarding information to reach much smaller files. Formats like JPEG and lossy WebP target detail the eye barely notices, but aggressive settings cause artifacts. Because the data is gone forever, re-saving lossy files compounds damage; keep a clean original.
Luminance
The perceived brightness of a pixel, independent of its color. Pipelines separate it from chroma because eyes resolve brightness far better than color, enabling chroma subsampling. It is also the quantity measured in accessibility contrast ratios.
M
Magic bytes
The identifying byte sequence at the start of a file that declares its format, also called its file signature. Browsers and operating systems read it to recognize files regardless of their extension. Detecting corrupted or mislabeled files relies on checking these bytes.
Metadata
Structured data inside an image file that describes the picture rather than its pixels. It ranges from camera settings and timestamps to keywords, copyright, and GPS. It powers search and management but can leak private details, so strip it before public sharing.
MIME type
The standardized label identifying a file’s kind, such as image/png or image/jpeg. Servers send it in response headers so browsers know how to render content. A mismatch explains why some images download instead of displaying inline.
Monochrome
An image in a single hue, usually black and white. True grayscale holds no color, while a tinted monochrome keeps one hue at varying brightness, as in sepia. Single-channel data lowers file size and can simplify a subject.
N
Nearest-neighbor interpolation
The simplest resampling method, copying the closest source pixel into each target pixel. It produces hard edges and pixelation when enlarging, so it is poor for photos. It is perfect for pixel art, where every chunky pixel must stay intact.
O
Optimization
Making an image as small as possible while keeping it acceptable for its purpose. It combines format choice, resizing to display size, a sensible quality setting, and metadata removal. Optimized images load faster, and faster pages rank and convert better.
P
Palette
The fixed set of colors an image can use, referenced by index instead of full color data. GIF holds a palette of up to 256 colors, keeping simple graphics small. Forcing photos onto a small palette causes posterization and color shifts.
Pixel
The smallest unit of a digital image, a single color and brightness sample. Raster images are grids of these samples, and quality depends on how many fit across each dimension. No algorithm can create detail the pixels never captured.
PNG
Portable Network Graphics, a lossless raster format built to replace GIF. It offers full alpha transparency, millions of colors, and excellent compression of flat or text-heavy graphics. Files are larger than lossy photographic formats, but pixels stay exact.
PPI
Pixels per inch, describing how many pixels of a file fit in one physical inch. It is the file-side counterpart of DPI and has no effect on screen rendering. A 72 PPI and a 300 PPI file hold the same pixels; only print size changes.
Progressive JPEG
A JPEG variant storing multiple passes so browsers show a rough full image first and refine it as data arrives. Viewers perceive this as faster, especially on slow connections. Files may be slightly larger but the loading feel usually wins.
Q
Quality factor
The numeric setting, usually 1 to 100, telling a compressor how aggressively to discard data. Higher values keep more detail and larger files. Most web work starts near 80 for photos and adjusts based on size budgets.
R
Resolution
The number of pixels an image holds, stated as width by height. Higher resolution means finer detail up to the source’s limit, plus larger files. It is a property of the file, not of the screen displaying it.
RGB
Red, green, and blue, the three additive channels screens combine into every visible color. Varying their intensities covers the full display gamut, from black at all-off to white at full. RGB is the color model of every digital screen.
RGBA
The RGB model extended with a fourth alpha channel for opacity. Storing alpha per pixel is what lets PNG and WebP show soft edges and transparency. Without it, those effects would need a solid color fill to fake them.
S
Saturation
The intensity of color, from muted gray to vivid. Reducing it moves an image toward grayscale; raising it intensifies hues until they clip. It is independent of brightness in models like HSL, so you can boost color without lightening the picture.
SVG
Scalable Vector Graphics, an XML format describing images as mathematical shapes rather than pixels. Because shapes are resolution-independent, an SVG stays crisp at any size and can be tiny. Ideal for logos, icons, and charts, though not for photographic textures.
T
Thumbnail
A small preview of an image used to browse libraries quickly and load galleries fast. Thumbnails stay under a few hundred pixels and a few dozen kilobytes. Generating them once at build time avoids downloading full images just for previews.
TIFF
Tagged Image File Format, a flexible raster format used in print and publishing. It supports high bit depths, layers, and rich metadata, but files are large. It is a working and archival format, not a web delivery format.
Transparency
The property letting parts of an image reveal whatever sits behind them. It is stored in an alpha channel and lives in PNG, WebP, and AVIF but not standard JPEG. Essential for logos, overlays, and cutouts on unknown backgrounds.
U
Upsampling
Increasing an image’s pixel count by interpolating new pixels between existing ones. Upsampling never adds real detail; it spreads existing information further and softens edges. Also called upscaling, it is best avoided when larger source files are available.
Upscaling
Enlarging an image beyond its original dimensions by interpolation or AI enhancement. Plain methods give soft, blurry results because they fabricate pixels from neighbors. AI upscalers invent convincing detail, making them a creative tool rather than a recovery tool.
V
Vector image
An image built from geometric shapes and curves rather than a pixel grid. It scales to any size without losing crispness and stays easy to edit. Formats like SVG make vectors ideal for logos and illustrations, while photos remain raster territory.
W
WebP
A Google-designed web format supporting lossy and lossless compression, transparency, and animation. Lossy WebP typically beats JPEG by 25 to 35 percent at equal quality. Support is essentially universal now, making it the safest modernization step for most sites.
White balance
The correction removing the color cast produced by the light source so neutral surfaces appear white. Cameras and editors rescale color channels to drop the temperature tint. Automatic balancing reads the scene, while manual settings are dependable in mixed light.
Wide gamut
A display or color space reproducing more colors than standard sRGB. Spaces like Display P3 hold richer greens, reds, and blues, which matters for HDR and photography. Viewing wide-gamut files on an sRGB screen compresses those colors, so profiles and conversion matter.
X
XMP
Extensible Metadata Platform, Adobe’s XML standard for descriptive metadata inside image files. Its fields cover creator, title, keywords, copyright, and edit history, traveling with the file across applications. It complements EXIF and IPTC in formats like JPEG, PNG, and TIFF.
Y
YCbCr
A color space splitting an image into luma (Y) and two chroma components, blue-difference (Cb) and red-difference (Cr). It is the working space inside JPEG and video where chroma subsampling happens. Converting RGB to YCbCr is the first step a JPEG encoder takes.
Keep this glossary open alongside the tools. When compression looks off, check lossy compression, quality factor, and banding. When a photo reveals more than intended, revisit EXIF and XMP. Two golden rules run through every entry: always work from the best-quality original you have, and choose the format that matches the content rather than the one you are used to.