In a 3D simulation, a real 3D model does what the narration says, on the word: a heart pumps, a jet engine turns, the Moon crosses the Sun. Until now the camera eased in and then held one shot for 10 to 15 seconds. That is a long time on a phone.
A new angle on every moment. Each time the narration starts a new moment, the camera moves to a new angle around the subject: a little round, a little higher or lower, a little closer. The move is quick at first, then settles. If more than 4.5 seconds pass without a new moment, it finds another angle, and it drifts slowly in between. Flat diagrams, such as the compound growth chart, turn only gently, so they never end up edge-on.
A fast opening. Every simulation now flies in to its first shot in the first couple of seconds, from farther out and from the side. The first seconds decide whether people keep watching.
The voice you pick. The narrator you choose in the idea box is now the one you hear. Before, the writer put in the language's default voice.
The same studio. The simulation studio now has the video studio's layout: the steps across the top, a Style card with the voice, and the script with a list of its moments and the word each one lands on.
Urdu, Arabic and Hindi captions. Captions and titles now use Nastaliq, Naskh and Devanagari fonts. On our render machines they had come out as empty boxes.