Prompt recipe makes AI video reveal title letters one by one in cinematic storm scenes

LudovicCreator · x · 2026-09-18

LudovicCreator shared a reusable prompt template for cinematic title animations: for the word "SURGE", letters must appear strictly one at a time (S, U, R, G, E), never showing the full word upfront, while the storm environment — rolling thunderclouds, layered rain, drifting mist, distant lightning, wet reflective ground — stays alive throughout. He validated the template on "FORGE" to show transferability. The key trick is explicit temporal sequencing plus negative constraints (no simultaneous reveals) to keep video models stable at rendering text frame by frame.

Original post →

More from Multimodal

Multimodal channel →