The most efficient way to add right-to-left languages to a left-to-right video pipeline is to split the job in two. Getting the text to appear in the right script is one problem. Getting the design to read correctly for someone who reads from the right is a different problem. The first is mostly a data and font question. The second is a design question, and a tool will not answer it for you.
This article is for the producer or localization lead whose template, spreadsheet, and review process were all built around English, and who has just been asked to deliver Arabic and Hebrew versions. We will be specific about what Dataclay’s own documentation says about right-to-left text, and equally specific about what it does not say, because a guess here ends with a client finding a backwards headline on air.
What our pages say, and what they leave open
Here is the full extent of what Dataclay documents about right-to-left languages in Templater, our After Effects plugin for data-driven video. On the Templater page, under the feature called “Use international languages,” we say that you can insert characters from any language into your spreadsheet and see them appear in your text layers. We add that installed fonts should have the glyphs you require, and that Templater reads your text as it appears in your spreadsheet, so commas, quotes, apostrophes, and new lines are accepted. Then we say right-to-left scripts are also supported. In a post about automated news video, we repeat that Templater supports diverse languages and right-to-left scripts. In a post about generative AI and media automation, we note that the dynamic text styling feature allows font-face and direction properties to be controlled directly from a data source.
That is the sourced list. Text in right-to-left scripts goes from the sheet into text layers. The font has to contain the glyphs. Direction is a property you can drive from the data.
Here is what that list does not tell you, and it is worth reading slowly. Our pages do not describe how a line that mixes right-to-left and left-to-right text, such as an Arabic sentence with an English brand name inside it, is ordered. They do not describe how numerals behave. They do not say how punctuation sits at the end of a line. They do not say how a layout rule treats an element that should mirror to the opposite side of the frame. They do not publish a list of languages that have been tested, and they do not say how any specific font shapes connected scripts.
So the honest summary is this. Templater is documented to carry right-to-left text and to let direction be set from data. Everything that happens after that, the shaping, the mixed-direction ordering, and the mirrored design, is something you have to verify on your own files with your own fonts. We would rather say that than promise a result our pages do not describe. If your project has a hard requirement in one of these areas, run a small proof before you commit, or contact us and tell us exactly what you need.
The sequence that wastes the least time
Teams lose time on right-to-left work because they start where it looks hardest, with the animation, and skip the dull checks that decide whether the animation will work. The order below goes from cheapest to most expensive, and each step either clears the next one or stops you early. This is general practice, built from how the sourced features fit together.
Step one: prove the text, with no design attached. Before you touch your real template, build a throwaway composition with one text layer, driven by one row of your data source. Put in a real sentence in the target language, supplied by whoever will be responsible for the translation, not a placeholder you typed or copied from a search result. Choose the font you intend to use. Run the merge and look at the text layer. You are checking three things. Do the characters appear at all, or do you get empty boxes where the font has no glyphs? Do the letters connect and run in the correct direction? Does punctuation land on the correct side? A person who reads the language will know in three seconds. A person who does not will not, and this is the moment to be honest with yourself about that.
This step costs almost nothing, and it settles the question that every later step depends on. If the font fails here, no amount of layout work will help.
Step two: decide who owns the font. Templater reads the text from your spreadsheet as it appears, and the font in the template must hold the glyphs. If your brand font does not cover the script, then the right-to-left template will use a different font, and that is a brand decision that belongs to the brand owner. Raise it early, in writing, with the font in question shown at the intended size. Do not make it yourself and show them the result at the end. If you render on several machines, remember that each machine needs the font installed. After Effects fails a render item when it cannot find a font dependency, and QUE satellites need the fonts installed before you assign them a job.
Step three: test the mixed lines. Real marketing copy is rarely pure. It carries product names in Latin letters, model numbers, percentages, URLs, and hashtags. This is where right-to-left text most often goes wrong in any tool, because each of those fragments has its own direction and the line has to order them correctly. Our pages do not document this behavior, so treat it as an open question until a proof settles it. Write the five ugliest lines from the real script, the ones with a brand name in the middle and a number at the end, and merge them in. Show them to a reader of the language. If the order is wrong, you have learned it for the price of a test and not a delivery.
Step four: separate what is data from what is design. Now decide, for every element in the template, whether it mirrors. This is the part that no feature does for you, and it is where judgment matters. Text direction is one question. Whether the whole composition reads from the right is another. A lower third with the logo on the left and the name to its right may need to flip so that the eye starts at the right edge. An arrow that points forward may need to point the other way. A progress bar that fills from left to right may need to fill from right to left. But a clock face, a video’s playback controls, and most logos do not mirror, and a flipped logo is a brand violation. Write a short list of the elements and what each does. Get the brand owner to sign it, ideally with a native reader looking at the list.
Step five: build the second template on purpose. Because layout rules work relative to the containing composition and sibling layers, a layout designed to read from the left can rarely just be told to read from the right. Our pages do not describe an automatic mirror. The honest, efficient approach is to build a second version of the template with the arrangement flipped where it should be, and keep it as a deliberate sibling of the first. Templater can switch project templates on a per-row basis from the same data source. That means a single sheet can hold your English, Arabic, and Hebrew rows, and each row can name the template it uses. You maintain two templates instead of one, and you keep one data source instead of three.
That is more work than a mirror button would be. It is also work that stays visible, reviewable, and under your control, and it avoids the situation where an automatic flip does something to a graphic that nobody checked.
A concrete example
Consider a producer at a financial services company who has a thirty second explainer about opening an account. The English version has a title card, three numbered steps with short captions, and an end card with a web address and the brand name. The company wants Arabic and Hebrew versions for two regional offices, and has a single reviewer who reads Hebrew and nobody in the building who reads Arabic.
The producer runs the sequence above. The proof composition shows that the brand font has no glyphs for either script, so the first deliverable is not a video at all. It is a note to the brand owner with two candidate fonts shown on a real sentence from each language. That conversation happens in a day, not in the last week.
The mixed-line test finds a problem the producer would not have found by looking. The end card carries the brand name in Latin letters and a web address inside a right-to-left sentence, and in the first merge the order of the fragments is wrong. The producer does not try to fix this in After Effects by hand. The sentence goes back to the translator with a screenshot, and the translator reorders the sentence in the source so that the fragments sit where they read correctly. The fix lives in the sheet, so it survives the next update.
The layout decision produces a short list. The three numbered steps run down the left in English and will run down the right in the other languages. The step numbers keep their numerals. The brand logo does not mirror. The arrow between steps flips. The producer builds a second template with this arrangement and adds a column to the sheet that names which template each row uses. The sheet has one row per language per caption, and a single data source drives every version.
Nothing about this is exotic. It is the same discipline that applies to any language batch, applied earlier and with more care at the points where the direction changes.
Checking the result when you cannot read it
The hardest part of right-to-left work for many teams is review, because the people building the pipeline often cannot read the output. A few habits help, and the first is to move the checking onto stills. In Templater Pro or Bot, spot checks export still images or short clips of selected frames from data-merged compositions. You name timeline markers to choose the frames, or Templater samples evenly spaced frames. For this work, put a named marker where the text is fully on screen, so every row produces one image at that moment. A reviewer can then go through a folder of stills quickly. They do not need to scrub video, and the review fits in a short call.
The second habit is to keep the reviewer’s job narrow. A reader of the language is not a layout reviewer, and a designer is not a language reviewer. Ask the reader a short list of questions on each still: is the order right, is any word broken or cut, do the letters connect, is the punctuation on the correct side. Ask the designer whether the element positions follow the signed-off list. When one person does both jobs they tend to do one of them badly.
The third habit is to keep the translated copy where it can be reviewed without the video. Templater reads from a spreadsheet, a tab-separated file, or JSON. If the text lives in a sheet, a reader of the language can review the words in one pass without any of the layout noise, and the stills only need to answer layout questions. Our pages document that Google Sheets can also draw in translations from Google’s translation service, but machine output in a language nobody on your team reads is not a deliverable. Treat it as a draft for a reader to correct.
The fourth habit is the one that saves the most heartache. Do not accept a batch because the first three rows look right. Right-to-left failures are often specific to certain strings, such as a line with a number at the beginning or a name in the middle, so a handful of clean rows says little about the rest. Choose the rows for review by what they contain, with the ugliest first.
If a row fails and a person has to correct a single version by hand, use Replicate. It inserts your data and saves a copy of the project for each version in an organized folder, so a designer can adjust the one that needs attention and render it from there. Keep this for exceptions. If you find yourself hand finishing a quarter of the batch, the template needs another pass, not the rows.
The bottom line
The efficient path is not a clever setting. It is an order of work. Prove the text and the font first, because everything else depends on them. Test the mixed-direction lines early, because that is where surprises hide, and our pages leave that behavior for you to verify. Decide element by element what mirrors, and get that signed. Build the right-to-left template as a deliberate second version and let the data source choose it row by row. Then review stills with a reader of the language and a designer, each answering their own questions.
What Dataclay documents is that right-to-left scripts are supported, that the font needs the glyphs, and that direction can be driven from data. What you should assume, until you have tested it, is that everything else about your particular design needs a proof. That is a modest claim, and it is the one we can stand behind.
For the wider picture, read our post on breaking language barriers with Templater, the walkthrough of spot checks for quality assurance, and the explanation of layout rules. The product itself is on the Templater page, and subtitles for the same languages are covered in dynamic subtitles in After Effects. If your sheet is the sticking point, see our list of tested data sources.
Here are the questions we hear most from teams adding right-to-left languages for the first time.
Frequently Asked Questions
Does Templater support right-to-left languages such as Arabic and Hebrew?
Our Templater page says that you can insert characters from any language into your spreadsheet and see them appear in your text layers, and that right-to-left scripts are also supported. Installed fonts need to contain the glyphs you require. We do not publish a tested language list or describe mixed-direction ordering, so test your own strings and your own font before committing to a delivery.
Can text direction be controlled from my spreadsheet?
Our post on generative AI and media automation says that Templater’s dynamic text styling feature allows font-face and direction properties to be controlled directly from a data source. What that covers in your specific design is something to confirm with a small proof, because we do not describe every case. Build one row, one text layer, and one real sentence in the target language and see what the merge produces.
Will my layout mirror automatically for right-to-left languages?
Our pages do not describe an automatic mirror. Layout rules adjust dynamic layers relative to their containing composition and sibling layers. The practical approach is to build a second template with the arrangement flipped where it should flip, and let your data source name the template per row, since Templater can switch project templates from data on a per-row basis.
What happens with a brand name or number inside a right-to-left sentence?
Our pages do not document how mixed-direction lines are ordered, so treat it as something to test. Write the five hardest lines from your real script, with a Latin brand name in the middle and a number at the end, merge them, and have a reader of the language look at them. If the order is wrong, ask the translator to restructure the sentence in the source so the fix lives in the data.
Do I need a different font for Arabic or Hebrew?
If your brand font does not include the glyphs, yes. Templater reads your text as it appears in the sheet, and the installed font has to contain the characters, or you will see empty boxes. Choosing a different font is a brand decision, so show the owner candidate fonts on a real sentence early. Remember every rendering machine needs the font installed too.
How do I review right-to-left output if nobody on my team reads it?
Move the review to stills. In Templater Pro or Bot, spot checks export images of frames you choose with named timeline markers. Send a reader of the language a folder of stills and a short question list about order, broken words, and punctuation. Keep translated copy in the sheet so the words can be reviewed without the video, and treat machine translation as a draft only.
