Video automation glossary

Making video

These terms describe how a video is made when one design is used many times: the template, the data that fills it and the render that turns both into a file.

Programmatic video

Programmatic video is video produced by running a coded template against structured data, so one template renders many videos. The template fixes layout, type, motion and timing, and each row, page or record supplies the values that change. Videngine is a programmatic video engine: it turns a web page, a listing, a block of text or a row of data into a narrated, captioned MP4. Read the full definition of programmatic video.

Render

A render is the process of computing every frame of a video from its template, data, images and audio, then encoding those frames into a finished file such as an MP4. The word is also used for the output: one render is one finished file. Renders can fail or need repeating, so a bulk workflow tracks each render job against its source.

Coded template

A coded template is a video template written as software, for example React components rendered with Remotion, that decides layout, timing and behaviour from the data it receives. Because it is code, it can handle cases an editor template cannot, such as a missing photo, a very long name or a price that should be hidden. Compare it with the other kinds in our guide to video templates.

Brand tokens

Brand tokens are the named values a template reads for a brand's typefaces, colours, logo, spacing, safe areas and motion. Setting them once means every video follows the same rules, and a brand change is made in one place rather than in every file.

Editor template

An editor template is a pre-designed video layout that a person opens in an online editor, fills by replacing the text and pictures, and exports one video at a time. It suits designed one-offs and seasonal pieces; at volume, every video is another manual session. See editor, API and data-driven templates compared.

Template API

A template API is a rendering service that fills a stored template with data sent by your own code and returns a video file. It gives a developer full control of the template, while the script, voice, captions, review and retries around each render remain the developer's to build. See what to evaluate before integrating a video automation API.

Scene plan

A scene plan is the ordered list of a video's scenes, with each scene's timing, message, on-screen text and the picture that supports it. It is the brief the template works from, and the cheapest place to catch a missing fact or an unsupported claim before anything is rendered. Work through an example scene plan for one product.

Batch run

A batch run is a single job that renders many videos from a list of inputs, usually one video per spreadsheet row, record or page. It needs consistent input data, a way to retry failures and a record that ties each output back to its source. The steps are set out in bulk video creation from a CSV.

AI video

AI video is a broad label: these terms separate tools that generate new footage from tools that assemble a video from words and pictures you already have.

AI video generator

An AI video generator is a tool that uses a generative model to create new video frames from a prompt, an image or other input, rather than assembling existing footage. The same prompt can produce different results each time, which suits concept and mood shots and makes exact facts harder to control. The term is also used loosely for any video tool with AI in its workflow; our guide to choosing an AI video generator sorts them by job.

Text-to-video

Text-to-video is the production of a video from written input. It covers two different jobs: a generative model creating footage from a prompt, and a tool turning a script into narration, scenes and captions. Text to video AI: prompt generators vs script to video explains how to tell them apart.

Image-to-video

Image-to-video is generative AI that takes a still image and creates new motion from it, such as a moving camera or an animated subject. It differs from a photo video, which moves across the real photographs with pans and zooms and invents nothing. Both are explained in photos to video: the two kinds.

Script-to-video

Script-to-video is a form of text-to-video in which your words become the narration, word for word, and the tool assembles visuals and captions for each scene. Choose it when the wording must stay exactly as approved. See how to write an explainer script for the ear.

URL-to-video

URL-to-video is a workflow that reads a web page, extracts its text, images and structured data, and turns them into a script and scenes. The result can only be as accurate as the page, so extracted facts need checking before rendering. Read how to turn a webpage into a useful video.

Avatar video

An avatar video is a video in which a digital presenter, a synthetic or digitally recreated person, speaks a script to camera. It suits training and internal communication where a face helps, and is a different job from showing a product, property or listing. See how avatar presenter tools compare with template rendering.

Frames and formats

A video's frame decides where it can be shown well: these terms cover shape, resolution, motion and the still image that stands in for the video.

Aspect ratio

Aspect ratio is the proportion of a video frame's width to its height, written as width:height. One source usually needs more than one ratio, because a placement crops or letterboxes a frame that does not match it. See which aspect ratio suits which placement.

16:9

16:9 is the landscape aspect ratio, sixteen units wide to nine high, used by most television, desktop players and long-form YouTube video. At 1080p it is 1920 × 1080 pixels. It is the usual frame for YouTube videos made from your pages.

9:16

9:16 is the vertical aspect ratio, nine units wide to sixteen high, that fills an upright phone screen. It is the usual frame for YouTube Shorts, Reels and TikTok; at 1080p it is 1080 × 1920 pixels. Landscape photos need reframing to fit it, as shown in framing photos for 9:16 and 16:9.

1:1

1:1 is the square aspect ratio, equal in width and height. It is used for feed posts and for placements where one file has to work on both desktop and phone; at 1080p it is 1080 × 1080 pixels.

4:5

4:5 is a portrait aspect ratio, four units wide to five high, used mainly for feed posts viewed on phones. It takes more of the screen than 1:1 without needing a full vertical frame; at 1,080 pixels wide it is 1080 × 1350.

Safe area

A safe area is the part of the frame where text and key subjects are expected to stay visible, clear of edges that may be cropped and of the interface a platform lays over the video. On vertical platforms the account name, post caption and buttons cover the bottom and right edge, so text placed there can be hidden. See what short-form video needs that landscape video does not.

Frame rate

Frame rate is the number of still images, or frames, shown per second of video, written as fps. Common rates are 24, 25, 30 and 60 fps; a higher rate gives smoother motion but means more frames to render and a larger file at the same quality.

1080p

1080p is a video resolution 1,080 pixels tall with progressive scan, which in 16:9 is 1920 × 1080 pixels and is also called Full HD. The same label is commonly used for vertical 1080 × 1920 video. It is the usual standard for web and social video; higher resolutions cost more render time, storage and bandwidth.

Poster frame

A poster frame is the still image shown in place of a video before it plays, set on a web page with the poster attribute of the HTML video element. Search engines use a video's thumbnail in its listing, and Google's video best practices say a video must have a valid thumbnail image to be eligible for video features. The thumbnail is also a field in VideoObject schema.

Files and platforms

A finished video is a file with a container and a codec, sent to a platform with its own rules; these terms cover that last step.

MP4

MP4 (MPEG-4 Part 14) is a container file format that holds video, audio and subtitle streams together with their timing. It is the most widely supported format for web, social and download delivery; the container says nothing about quality, which depends on the codec and bitrate inside it.

WebM

WebM is an open container format for web video, usually carrying VP8, VP9 or AV1 video with Vorbis or Opus audio. Most modern browsers play it, and it is often offered beside an MP4 as an alternative source for the same video.

H.264

H.264, also called AVC (Advanced Video Coding), is a video compression codec that stores video at a small fraction of its raw size. It is the most widely compatible codec for MP4 files, playable on nearly every browser, phone and video platform.

Bitrate

Bitrate is the amount of data used per second of video or audio, usually measured in kilobits or megabits per second (kbps or Mbps). At the same resolution and codec, a higher bitrate generally means fewer compression artefacts and a larger file; most platforms re-encode uploads to their own bitrates anyway.

YouTube Shorts

YouTube Shorts are short square or vertical videos on YouTube, shown in their own swipeable feed. For standard channels, YouTube classifies uploads made after 15 October 2024 with a square or vertical aspect ratio and a length of up to three minutes as Shorts. See Shorts, Reels and TikTok at scale.

Reels

Reels are Instagram's short-form vertical video format, shown full screen in their own swipeable feed. Length limits and recommendations change, so check Instagram's own guidance before planning to a maximum; our short-form guide keeps a dated summary.

Captions and audio

Narration and captions carry a video's words twice, as speech and as text; these terms cover how the speech is produced and how the text is attached.

Voiceover

A voiceover is narration recorded separately from the pictures and laid over them, by a human speaker or a synthetic voice. In template-built video it usually carries the facts, so the on-screen text can stay short. See our guide to AI voiceover.

Synthetic voice

A synthetic voice is speech generated by a text-to-speech model from written text, with no person recording the line. A corrected line can be regenerated without booking a new recording session, and the result depends as much on the script and pronunciation rules as on the model. Choosing and directing an AI voice covers the trade-offs.

Voice cloning

Voice cloning is the creation of a synthetic voice modelled on recordings of a specific real person. It needs that person's informed consent and clear terms on where the voice may be used, and it is distinct from choosing a stock synthetic voice or uploading audio you recorded yourself.

Pronunciation lexicon

A pronunciation lexicon is a list that tells a speech engine how to say specific words, such as brand names, model numbers, place names and acronyms. Set once per brand, it keeps the hundredth video saying a name the same way as the first. See a pronunciation lexicon for product names, acronyms and API terms.

Transcript

A transcript is the full text of what is said in a video, without timings. Published on the page as readable text, it serves people who prefer reading and gives search engines and AI systems the spoken content as text. See transcripts and captions for video SEO.

Closed captions

Closed captions are timed text tracks delivered alongside a video that the viewer can turn on or off. They live in a separate file, such as WebVTT or SRT, which the player displays and platforms can read. Strictly, captions also describe meaningful non-speech sound, which is what distinguishes them from subtitles.

Burned-in captions

Burned-in captions, also called open captions, are captions rendered into the video frames themselves, so they always show and cannot be turned off. They survive any player or platform that ignores caption files, but a correction means a new render, and players and search engines cannot read them as text. They are standard in short-form video watched with the sound off.

SRT

SRT (SubRip Subtitle) is a plain-text caption file format made of numbered cues, each with a start time, an end time and the text to show. It is widely accepted for uploading captions to video platforms and editors, but its support for styling and position is informal and varies by player.

WebVTT

WebVTT (Web Video Text Tracks) is the caption format used by the HTML video element, loaded through a track element. It resembles SRT but adds a file header, cue settings for position and alignment, and support for chapters and metadata tracks. It is the caption track to publish with a video on your own site.

Video SEO and AI search

Search engines and AI assistants can only show a video they can find, fetch and understand; these terms cover the page, the markup and the files that make that possible. The video SEO guides cover each one in depth.

Watch page

A watch page is a page whose main purpose is to show a single video, so watching it is the main reason someone visits. Google's video SEO best practices use the term and distinguish watch pages from pages where a video complements the text. Where your watch pages should live is covered in YouTube or self-hosted video.

Structured data

Structured data is a standardised format for providing information about a page and classifying its content, usually written with the schema.org vocabulary. Google Search supports it in JSON-LD, Microdata and RDFa; it can make a page eligible for rich results, but Google does not guarantee that markup will produce a particular feature. For video, the type to use is VideoObject.

JSON-LD

JSON-LD (JavaScript Object Notation for Linked Data) is a way of writing structured data as a JSON object inside a script element, kept separate from the visible HTML. Google recommends JSON-LD for structured data where a site's setup allows it.

VideoObject

VideoObject is the schema.org type that describes a video: its title, description, thumbnail, upload date, duration and file or player location. Google's video structured data documentation lists name, thumbnailUrl and uploadDate as required, and recommends contentUrl, embedUrl, duration and description among others. See how to add VideoObject schema to a video page.

Rich result

A rich result is a search result shown with more than the standard title, link and snippet, such as a video thumbnail, key moments or a LIVE badge. Valid structured data can make a page eligible for one, but the search engine decides whether to show it.

Key moments

Key moments is a Google Search feature that lets people jump to segments of a video, like chapters in a book. A site can label the segments itself with Clip markup, or use SeekToAction markup to show Google where timestamps go in its URLs so Google can identify the moments automatically.

Video sitemap

A video sitemap is a sitemap with additional information about the videos hosted on a site's pages. Google's video sitemap documentation requires a thumbnail, title and description for each video, plus either the video file's location or a player location. See how to build and maintain a video sitemap.

robots.txt

robots.txt is a plain-text file at the root of a site's host that tells crawlers which URLs they may access. It manages crawler traffic rather than keeping pages out of search, and Google's video guidance says not to block video files or thumbnails with it if the video should appear in results. The file can also list the location of a sitemap; video in AI search covers rules for AI crawlers.

Crawler

A crawler is a program that automatically discovers and scans websites, following links and fetching pages; it is also called a bot or spider. Search engines and AI companies run their own crawlers, and most identify themselves with a user-agent name that robots.txt rules can address.

AI Overviews

AI Overviews are AI-generated summaries that Google Search shows for some queries, with links to supporting web pages. Google's guidance on AI features says there are no additional requirements or special markup to appear in them: a page must be indexed and eligible to be shown with a snippet. See how video appears in AI search.

GEO (generative engine optimisation)

GEO (generative engine optimisation) is the practice of making content easy for AI answer engines, such as Google's AI Overviews and AI assistants that search the web, to find, understand and cite. In practice it overlaps heavily with ordinary SEO: crawlable pages, direct answers, accurate structured data and text versions of what videos say. Video in AI search applies it to video pages.

llms.txt

llms.txt is a proposed convention for a Markdown file, usually placed at a site's root as /llms.txt, that gives language models a short summary of the site and links to its most useful pages. It is a community proposal rather than an adopted standard, and Google says no special AI text files are needed to appear in its AI features.

Workflow and review

Producing many videos turns quality into a process; these terms cover how videos are checked, recorded and costed.

Review and approval

Review and approval is the step in which a person checks each finished video against its source and brand rules, then approves it for publishing or rejects it with a reason. It is the control that stops automated volume becoming automated mistakes, so it needs written acceptance criteria and, ideally, a reviewer other than the person who built the video. See how to run video production with approvals built in.

Defect code

A defect code is a short, fixed label recorded when a video is rejected, naming the type of fault, such as pronunciation, photo or brand. Counting rejections by code shows which fix helps most: a lexicon entry, a template change or better source data. See defect codes at work in a newsroom video workflow.

Version

A version is one saved render of a video; a rebuild after a correction becomes a new version, such as _v2, kept beside the original rather than overwriting it. Keeping versions shows what was published and when, which matters if a price, specification or claim is later questioned. See how versions work for dealer listing videos.

Video ledger

A video ledger is a record of every video produced: its source, versions, review status, builder, reviewer, any defect code and where it was published. Joined to a listing or product table, it shows which live items still lack an approved video, as in marketplace listing video.

Cost per approved video

Cost per approved video is the total cost of production, including tool fees, extra usage and review time, divided by the number of videos that meet your acceptance criteria. It is a fairer comparison than the price per export, because rejected attempts and revisions are counted. Work it through in AI video generator costs.

Choose your next step

Use these definitions with the two guides that put them in order: one for choosing a tool, one for getting the result found.

• Choose an AI video tool by the job it has to do
• Make your videos findable in search and AI answers