The demo is a 109-word script read in 43.9 seconds, about 2.5 words per second, over a cinematic look: clocks floating in the air, a planet under cloud, a man in a long green coat, a neon city in the rain. The mood is right. A Doctor who arrives “not with a roar, but with a whisper” is a very Doctor Who idea, and the closing question, stay or travel, invites real answers.
It also shows two traps. Listen to the third sentence: “In this faceless story, the Doctor arrives.” Faceless describes the video format, not the story, and the word leaked from the request into the narration. Read every AI draft once before you generate, because the narrator says exactly what is on the page.
The second trap is vagueness. Apart from the Doctor, nothing in the script comes from the show: no TARDIS, no companion, no monster. Lines like “a choice fractures gravity itself” sound grand but give the image model nothing specific, which is why the last scenes drift into a generic neon street with unreadable signs.
A sharper version keeps the length and names one thing, such as the TARDIS standing in that 1963 junkyard, or the first time a statue moves while nobody is looking. At this pace, about 110 words fills 45 seconds.
