LOADING
WHAT IT DOES
The AI is never allowed to draw a letter.
Gibberish lettering is the giveaway of a generated film, and it is not a quality problem — it is an architectural one. Video models cannot spell, so nothing here asks them to.
Words are composited, not generated
Titles, captions and your logo are laid over the film at assembly, the way an editor lays them on. They read as typed because they were typed.
- On-screen headlines held in their own field, never in the shot's prompt.
- Your logo placed as your file, not redrawn from a description.
- Captions burned in from real timings, in the style you pick.
It is enforced upstream
The scene graph keeps overlay text in `overlayHeadline` and `overlaySub`, separate from the visual prompt, precisely so a well-meaning rewrite cannot slip a word into the picture.