Broadcasters automate graphics from live data in After Effects by designing each graphic once, linking its changing parts to a spreadsheet or data feed, and letting Templater render a finished version for every row of data. The repeatable graphics belong to the templates. The judgment calls belong to people. The skill is knowing where one stops and the other starts, and building a hand-off that makes the line obvious.
Most graphics teams already feel this line, even if they have never drawn it. Nobody enjoys retyping a surname into the thirtieth lower third of the week. Nobody wants a machine deciding how to frame a breaking moment either. This article walks through both sides honestly, then shows how to design the point where the work passes from the data to a human pair of eyes.
What a graphics team can safely hand to templates and data
The test is simple. If a graphic has the same design every time and only its content changes, it can come from a template. If the content already exists in a sheet, a feed, or a form, it can come from data. When both are true, a person copying values into text layers is adding delay and risk, and no craft.
Lower thirds are the clearest case. A name, a title or a stat line, a bit of animation, and a placement that never moves. A sportsbook that streams live for twelve hours a day uses this approach for its lower thirds, which carry player names, statistics, or breaking news. The team fills in the information for the day in a Google Sheet and then performs a batch export when the sheet is complete. The same story reports that this dramatically reduced the time needed to produce titles. You can read the details in the full account of that workflow.
Stat graphics follow the same logic. A scoreboard card, a comparison of two players, a leaderboard, a season total. The design is fixed and the numbers are the only thing that moves. When the numbers live in a spreadsheet or a JSON feed, the graphic can update from them directly, because Templater reads Google Sheets, tab-separated files exported from a spreadsheet app, JSON files, and JSON feeds. Each row of the data becomes its own render job.
Schedule cards and promos are even more forgiving. The long-running stream above fills the gaps between its featured segments with graphics that promote current contests, show what is coming up next on the stream, and surface information from the sportsbook website. None of these needs an editor’s opinion. They need to be correct and on brand, and a template guarantees the second part while the data supplies the first.
Branding pieces belong here too. Stings, interstitials, and overlays are the glue of a channel’s identity, and they repeat constantly with small changes of text or color. Templater can customize network branding graphics such as these, and its dynamic text styling lets the font face and the text direction change based on the data. A television awards organization used the same idea at scale. Its graphics are driven by hundreds of award categories and the names of winners, and linking After Effects compositions to external data turned a slow manual process into an automated one.
Captions and translations are a quieter win. Templater can automate subtitles, closed captions, translations, and on-screen text, and it supports international languages and right-to-left scripts, provided the installed fonts have the glyphs you need. A person still has to know whether the translation reads well. The typesetting, timing, and placement of it can be handled by the template.
Social versions come last, and they are where the arithmetic gets dramatic. One lower third can become a vertical clip, a square clip, a still, and a version with different sponsor branding, each one a row in the sheet. Projector can then produce GIF clips and poster images along the way, so the social team is not rebuilding the same asset five times.
What still needs a human making judgment calls
Now the other side of the line. Here is what the data cannot know, and what a template should never be asked to decide.
Editorial judgment comes first. A sheet can say who scored. It cannot say whether the moment deserves a full-screen treatment, a quiet mention, or no graphic at all. Choosing what the audience should see next is a producer’s job, and it stays one. Templates make the chosen graphic cheap to produce. They do not choose it.
Breaking moments come second. Live broadcasting has a habit of producing something nobody planned for. A rule change, a delay, a sudden story that needs its own treatment. Your templates will cover the situations you designed for. They will not cover the one you did not, and a designer building a new graphic from scratch in the middle of a broadcast is still a person doing craft work under pressure. Automation helps most before and after that moment. It clears the routine graphics off the team’s plate so the people are free when the unplanned thing arrives.
Tone is third. The same data can call for a celebratory graphic or a restrained one, and a template carries only the tone it was designed with. An injury update and a championship celebration might share a data field and almost nothing else. A person has to look at the context and decide which template the row should use.
Final checks are fourth. This is the point the research on automated video makes clearest. A study of 4,200 UK news consumers compared videos that were made by humans, produced mostly by automation, and produced by automation and then edited by people. As we summarized in our post on why automated news video is better with a human touch, the study found that AI-generated videos for short messages are only as well received as manually created ones if they are edited by humans. The partly automated videos, which were post-edited by humans, were not liked significantly differently from the human-made ones overall. The highly automated videos were liked significantly less.
Read that result carefully, because it cuts both ways. It supports automation, since the edited, partly automated videos held their own against fully human-made ones. It also says that skipping the human step costs you with the audience. The study also found that post-editing did not eliminate every shortfall. Problems with the quality and variety of the images in the automated videos remained, and the study noted that the music chosen by the software’s operator was judged less suited to the story. Those are exactly the kinds of choices a template cannot make for you. The study’s takeaway was that automation works best as a hybrid, with human supervision supplying quality while the computer supplies speed and scale.
Finally, there is anything the data cannot know. A headshot cropped badly by the person who supplied it. A name spelled one way in the roster and another way on the broadcast. A stat that is technically correct but was revised at halftime. A sponsor that must not appear next to a certain story. The sheet will happily hand all of these to the template, and the template will render them faithfully. Faithful is not the same as right.
How to design the hand-off
The useful question is not whether a human is involved. It is where. A good workflow puts a person at a small number of deliberate points, instead of everywhere or nowhere.
Put a preview between the data and the render
Templater lets you preview a composition with your data before you render or replicate it. You can step through the render jobs and see how the template treats different text, footage, and colors, which makes problem data easy to resolve while it is still just a cell in a sheet. For sports and live events, this is the cheapest quality check available, because the longest surname and the missing photo show up in the preview, not on air.
Use spot checks when the batch is too big to watch
Previewing every row stops being realistic when a batch has hundreds of jobs. That is what spot checks are for. When you generate spot checks, Templater exports still images or movie clips of individual frames, or groups of frames, from the data-merged compositions. You choose the frames with named timeline markers on the composition, and if there are no markers, Templater samples an evenly distributed number of frames across the duration. You can choose the file format, and the path of each spot check can be passed to a script registered to a supported Templater event, which means the previews can be routed to whoever signs off.
Two further details matter to broadcast teams. Spot checks can verify how data controls properties promoted through Dynamic Essential Properties, available in Templater 3.5 or later. They can also verify the effect of Time Sculpting rules, the time-based rules that crop, trim, shift, and stretch layers so footage of different lengths fits the design. Spot checking needs Templater 3.0 or newer and is available in the Pro and Bot editions. Our post on generating spot checks walks through the setup.
Choose which points are automatic and which are not
Templater comes in three editions, and the choice quietly decides where your humans sit. Pro renders or replicates a batch when someone clicks. That click is a human-in-the-loop point built into the tool. Bot checks periodically for incoming data and starts renders on its own, or takes data pushed in from the command line, so it is meant for work that happens without a person at the keyboard.
For a broadcast team, this suggests a sensible mixture. Lower thirds for a routine segment might run through a Bot with a spot check reviewed on a schedule. A high-profile graphic, such as a season opener package, might go through Pro so that a person starts the batch after looking at the preview. Same templates, different supervision, chosen by how much a mistake would cost.
Keep a hand-finishing path open
Sometimes the automated version is ninety percent of the way there and a designer wants the last stretch. Templater can replicate each row into its own composition inside After Effects instead of rendering it right away, kept tidy in a clean folder structure. The designer can then adjust the few that need it and send everything to the render queue or to Adobe Media Encoder. The source describes this as reducing repetitive drudgery while allowing human interaction in the versioning process, which is the whole idea.
Fix mistakes in the data, not in the project
This is the most useful habit in the whole workflow. When something changes at the last minute, an injury, a schedule move, a corrected figure, the correction is an edit to the data and the graphic regenerates. The sportsbook team described above handles last-minute changes like unexpected player injuries or changes in match schedules this way, by updating the spreadsheet, and its lower thirds are updated frequently in real time during live events. Nobody reopens the After Effects project under pressure. The design stays untouched, so the correction cannot break it.
It also makes the roles clear. The designer owns the template. The person closest to the information owns the data. In that same stream, the customer experience team fills out the spreadsheet and the graphics come out in minutes instead of hours, so people without design experience get to use their in-house artists’ work without adding to anyone’s workload.
Wiring live data into the pipeline
A spreadsheet is the right starting point for most teams. It is easy to share, easy to read, and everyone already knows how to use one. Eventually, though, live graphics want data from places other than a person typing.
QUE is the next step. It is a cloud-based data source built for Templater that stores each job as a JSON object, which you can picture as a spreadsheet row, and you do not need to understand JSON to use it because you can enter information in the QUE Dash. It exposes API endpoints that can receive data from anywhere, including mobile, web, or voice apps, and its built-in Zapier integration connects to more than 5,000 apps without coding. It organizes work as Autografs, Jobs, Campaigns, and Satellites. A Campaign is a collection of Jobs, and Satellites are the computers running After Effects and Templater that process them. The dashboard shows each job’s status and the utilization rates of rendering machines, which gives a producer a place to see what is waiting and what is done.
A feed can also be a JSON URL. Templater reads JSON feeds directly, so a score source, a stats service, or a form can supply the rows. This is the plain answer to the question of how broadcasters automate graphics from live data. The data lives somewhere that updates, the template is linked to it, and the render step is either triggered by a person or watched for by a Bot.
Speed on the render side comes from Projector, which renders several customized videos at the same time based on the number of cores in the machine. An eight-core machine renders eight at once. It can also transcode as part of each render and deliver files to Vimeo, JW Player, Frame.io, and AWS S3, and it writes metadata from the streaming platform back to your data source. It runs locally on your own machine, which means your unreleased assets stay under your control.
Where this is the wrong tool
A few honest limits.
Templater is built around After Effects. If nobody on your staff or your agency works in it, someone will have to build the first templates. Dataclay offers custom development for teams in that position.
It is also not a replacement for a producer. Nothing in the tools decides which graphic the moment calls for. Teams that expect the software to make editorial choices will be disappointed, and teams that treat it as a fast, accurate assistant to their editorial choices tend to be happy.
And the preview habit takes discipline. Spot checks only protect you if someone looks at them. A stack of previews that nobody opens is the same as no quality control, and the hand-off falls apart quietly. Decide who reviews which output before the first live night, and write it down.
Bottom line
Hand templates and live data everything that repeats: lower thirds, stat graphics, schedule cards, promos, stings and interstitials, captions and translations, and every social version of each. Keep people on everything the data cannot know: what deserves airtime, how to handle a breaking moment, which tone fits, and whether the final frame is right. Then build the hand-off on purpose, with a preview before the batch, spot checks for the big runs, a hand-finishing path for the special ones, and corrections made in the data instead of the project. The evidence we have says audiences accept automated video when a human has edited it. Design for that and you get the speed without the risk.
The routine part should cost you very little. The part that needs a person should get all of their attention.
Frequently Asked Questions
Which broadcast graphics are safe to fully automate?
Graphics where the design never changes and only the content does. Lower thirds, stat cards, schedule cards, promos, stings, interstitials, and captioned or translated versions all fit. Templater links the changing parts to a Google Sheet, a tab-separated file, a JSON file, or a JSON feed, and each row of data becomes its own render job. Anything that depends on editorial judgment should still pass through a person.
Does automated video need a human to review it?
The research we cover says it helps a great deal. A study of 4,200 UK news consumers found that AI-generated videos for short messages are only as well received as manually created ones if they are edited by humans. Highly automated videos were liked significantly less. Templater supports review through previews, spot checks, and replication for hand-finishing.
What is a spot check in Templater?
A spot check is a still image or movie clip exported from specific frames of a data-merged composition, used to check a batch before or after rendering. You mark frames with named timeline markers, or Templater samples frames evenly across the duration. Spot checking needs Templater 3.0 or newer and is available in the Pro and Bot editions.
How do you handle a last-minute change on a live show?
Edit the data, not the After Effects project. If a player is injured or a schedule changes, you update the cell in the sheet and the graphic regenerates with the right information. A sportsbook that streams live for twelve hours a day handles changes this way, and its lower thirds are updated frequently in real time during live events.
Can a Bot run without anyone watching?
Yes. Templater Bot checks periodically for incoming data and starts new renders on its own, and it can also take data pushed in from the command line. The Pro edition renders or replicates a batch when someone clicks, which gives you a built-in human checkpoint. You can use both, depending on how costly a mistake would be.
Where does the live data come from?
Templater reads Google Sheets, tab-separated files, JSON files, and JSON feeds. For larger setups, QUE stores each job as a JSON object in the cloud and accepts data through its API or through Zapier, which connects to more than 5,000 apps. A person typing into a sheet, a form, or another system can all supply the rows.
