The Work
The Meta Quest 3 launch was a significant moment for Meta, their most capable mixed reality headset to date, and a product that needed trailers to match. Our work focused on the copy animations and visual effects across the official launch trailers, and then the full localisation of those trailers into over 12 languages for markets worldwide.
The source trailers set a high bar visually, and maintaining that quality through localisation, where copy length, text direction and character structure all vary, was the core challenge of the project.
Rotoscoping and Effects
One of the standout technical elements in the introduction trailer was getting the copy to pass behind objects in the scene. Specifically, the toys featured at around the 40-second mark. Achieving this required rotoscoping: frame-by-frame masking of the foreground elements so the text could be composited behind them. This had to be rebuilt or adapted for each localised version, since the copy length and line breaks changed with every language.
Localisation
Localising motion content is a different challenge from localising static copy. When text animates, such as fading in word by word, tracking across the screen, or timing to a beat, any change in word length, line count or reading direction can break the rhythm of the whole sequence.
German presented difficulties around copy length. Some translations ran significantly longer than the English source, which meant either the animation timing had to be extended, the copy had to be renegotiated with the localisation team, or sometimes even both. We worked closely with the localisation team throughout to find the right balance between translation accuracy and what was workable within the animation.
Korean was the most technically demanding. The intro trailer featured an animation of the word ‘play’ at around 15 seconds, which worked cleanly in Latin-script languages. In Korean, individual letters within a word are grouped into syllable blocks, so the character-level animation logic that worked in English didn’t translate directly. Another issue was that the word ‘play’ didn’t appear at the start of the sentence; it was placed more towards the end. We had to rethink the animation approach entirely for that market, building a version that respected how Hangul is read and rendered while preserving the intent of the original effect.
In the end, the simplest and most effective solution was to duplicate the animation, one where the individual letters or character blocks could be used when the letters were floating around, and then another where the correct word in Hangul flies into the sentence. We also adjusted the rest of the sentence, which didn’t animate as much in Latin-script languages, so that it animated seamlessly with the adjustments we made.
