Glossary

An image format developed by Google that supports both lossy and lossless compression plus transparency. At the same visual quality it is typically 25–35% smaller than JPEG, and unlike AVIF it is supported by every current browser. It is the safe default for web images when you also need older-browser support.

Used by: Image Converter, Image Compressor

A modern image format derived from the AV1 video codec. It produces the smallest files of any widely supported format — often half the size of JPEG at equal quality — and handles transparency and high dynamic range. The trade-off is slower encoding and no support in older browsers.

Used by: Image Converter

A measure of how similar two images look, scored from 0 to 1. Unlike a raw pixel difference it tracks structure, so it barely moves for a slight brightness shift but drops sharply when edges and texture are damaged — which is what compression does. Above 0.99 is visually identical; above 0.97 is indistinguishable in normal viewing.

Used by: Image Converter

Lossy compression permanently discards information the eye or ear is unlikely to notice, achieving much smaller files — JPEG, MP3 and H.264 all work this way. Lossless compression stores the data more efficiently without discarding anything, so the original can be reconstructed exactly. PNG, FLAC and ZIP are lossless.

Metadata that cameras and phones embed in a photo: the device, exposure settings, timestamp and often the exact GPS coordinates where it was taken. It travels with the file when you share it. Re-encoding an image through a converter drops this metadata, which is usually what you want before posting a photo publicly.

Used by: Image Converter

How many dots of ink or pixels are packed into an inch of printed output. It only means something in relation to a physical size: an image is not "300 DPI" on its own. Screens generally need 72–96 DPI and print needs around 300, which is why a web-sized logo looks soft when printed.

Used by: Passport Photo Maker

The proportional relationship between an image or video’s width and height, written as 16:9 or 1:1. Changing it without cropping stretches the content. Most social platforms crop uploads to a fixed ratio for each placement, which is why a correctly sized image can still appear cut off in a feed preview.

The algorithm that compresses and decompresses audio or video — H.264, VP9, AV1, AAC, Opus. It is distinct from the container that holds the result. A codec determines quality and file size; whether a device can play your file usually depends on whether it supports that specific codec.

Used by: Video Converter

The wrapper that holds compressed video, audio, subtitles and metadata together — MP4, WebM, MKV, MOV. It is not the same as the codec inside it, and containers accept different codecs: WebM cannot carry AAC audio, for example, which is the usual reason a converted file plays with no sound.

Used by: Video Converter

How much data is used per second of audio or video, measured in kbps or Mbps. It is the main lever on both quality and file size: file size is roughly bitrate multiplied by duration. Speech is transparent around 64–96 kbps, while music generally needs 128 kbps or more.

Used by: Audio Compressor, Video Compressor

A quality target for video encoding, where a lower number means better quality and a larger file. The scale is codec-specific rather than absolute — roughly CRF 26 on x264 matches CRF 30 on x265 — so switching codec at the same number silently changes the quality you get.

Used by: Video Compressor

Encoding a video twice: the first pass analyses the footage and records where detail and motion are concentrated, and the second pass spends the bitrate budget accordingly. It takes roughly twice as long but hits a target file size far more accurately than a single pass, which matters when an upload limit is fixed.

Used by: Video Compressor

A broadcast standard for measuring perceived loudness rather than raw peak level, expressed in LUFS. Normalising to it makes a whole file sit at one consistent loudness, which is why a normalised podcast does not force the listener to keep adjusting the volume between a quiet interview and a loud intro.

Used by: Audio Compressor

How many times per second an audio waveform is measured, in Hz or kHz. CD audio is 44.1 kHz and video audio is usually 48 kHz. A rate can capture frequencies up to half its value, so 8 kHz is enough for intelligible speech but removes everything that makes music sound full.

Used by: Audio Compressor

Mono carries one audio channel, stereo carries two with separate left and right content. Downmixing stereo speech to mono halves the file size with no perceptible loss, because a single voice carries no meaningful stereo information. Music does, so downmixing it flattens the spatial image.

Used by: Audio Compressor

Reading text out of an image or a scanned page so it becomes selectable, searchable and editable. A scanned PDF is a picture of a document and contains no text at all until OCR is run on it — which is why copying from one produces nothing and why conversion tools return empty output.

Used by: PDF OCR, OCR Image

The machine-readable text stored inside a PDF alongside its visual appearance. A PDF exported from Word has one; a PDF made by scanning paper or photographing a page does not. Whether a text layer exists determines whether a document can be searched, copied from or converted without OCR.

Used by: PDF Converter

A scanned PDF with an invisible OCR text layer placed behind the page image. It looks exactly like the original scan but can be searched, copied from and indexed. This is the standard way to archive paper documents without losing their visual fidelity.

Used by: PDF OCR

An ISO-standardised subset of PDF intended for long-term archiving. It requires all fonts to be embedded and forbids features that depend on external resources or that may not be renderable decades later, such as JavaScript and encryption. Many government and legal submission systems require it.

The dominant text encoding on the web, capable of representing every character in Unicode — Latin, Devanagari, Bengali, Arabic, emoji. Files saved in an older regional encoding and then read as UTF-8 produce question marks or mojibake, which is the usual cause of non-English text breaking during a conversion.

A low-level binary format that runs in the browser at close to native speed. It is what makes browser-based file processing practical: real image codecs, FFmpeg and PDF engines are compiled to WebAssembly and run on your own device, so a converter no longer needs to upload your file to a server to do the work.

Used by: How Farvixo works

Google’s three field-measured metrics for user experience: LCP for loading, INP for responsiveness and CLS for visual stability. They are measured from real visitors rather than a lab test, and they are a confirmed — though modest — ranking signal. Good targets are LCP under 2.5s, INP under 200ms and CLS under 0.1.

How long it takes for the largest visible element — usually a hero image or headline — to finish rendering. It approximates the moment a visitor feels the page has loaded. Under 2.5 seconds is considered good; the usual causes of a poor score are slow server response, render-blocking resources and unoptimised images.

How long the page takes to visibly respond to a user interaction, measured across the whole visit and reported near the worst case. It replaced First Input Delay in 2024 because FID measured only the delay before processing started, not how long the response actually took. Under 200 milliseconds is good.

A score for how much content jumps around while a page loads — the reason a mistap happens when a button moves just as you reach for it. Under 0.1 is good. It is caused almost entirely by images without declared dimensions, late-loading fonts and content injected above what is already visible.

A website that can be installed to a device’s home screen and work offline, using a service worker to cache its own files. It behaves much like a native app without an app-store download. Farvixo is installable this way, so its tools keep working on an unreliable connection.

The format Google recommends for structured data — a block of JSON embedded in a page that describes what the page is about in Schema.org vocabulary. It makes a page eligible for rich results such as FAQ accordions and breadcrumb trails, and it is how AI search engines identify entities reliably.

Used by: Schema Markup Generator

A declaration of which address is the authoritative version of a page, given as a link tag in the head. It stops the same content reachable at several URLs — with tracking parameters, with and without a trailing slash — from being treated as duplicates that compete with and dilute each other.

Structuring content so AI answer engines can extract and cite it — clear definitions, factual tables, explicit limitations and semantic markup, rather than keyword density. Where classic SEO competes for a click on a results page, GEO competes to be the source an assistant quotes when it answers without one.

A fixed-length fingerprint computed from data of any size. The same input always produces the same hash, and changing a single byte produces a completely different one. Hashes verify that a file or message has not been altered — they are not encryption, because the original cannot be recovered from them.

Used by: Hash Generator

A hash published alongside a download so you can confirm the copy you received is byte-for-byte identical to the one released. A mismatch means the file was corrupted in transit or tampered with. Use SHA-256 rather than MD5 for this — MD5 detects accidents but cannot detect deliberate tampering.

Used by: Hash Generator

A cryptographic hash function producing a 256-bit output, and the current default for verifying file integrity and authenticity. Unlike MD5 and SHA-1, no practical way is known to construct two different inputs with the same SHA-256 hash, which is what makes it trustworthy for security purposes.

Used by: Hash Generator

A compact, signed token carrying claims about a user or session, used widely for API authentication. Its payload is Base64-encoded, not encrypted — anyone holding the token can read its contents. That is why a live JWT should never be pasted into a website that sends it to a server to decode.

Used by: JWT Decoder

A way of representing binary data using only printable ASCII characters, so it can travel through systems that expect text — email attachments, data URIs, JSON payloads. It is an encoding, not encryption: anyone can decode it instantly, so it offers no protection whatsoever.

Used by: Base64 Encoder & Decoder

Doing the work in the visitor’s own browser instead of on a server. For a file tool this means the file is never uploaded: it is read, processed and saved locally. The privacy benefit is structural rather than a promise — a server that never receives your file cannot leak, retain or be compelled to hand it over.

Used by: How Farvixo works

Encrypting data so only the sender and intended recipient can read it, with no intermediate service holding a usable key. It is often confused with encryption in transit (HTTPS), which protects data on the wire but leaves it readable by the server at the other end.