I watched some, and I’m about as “pro AI” as you’ll find on Lemmy, but honestly:

• Tons of scenes are jarring. Like, right off the bat, the zoomed in scene of walking feet is token. Heads twist around, objects float; it’s creepy. It continues all through the movie, like the scene of the airplane landing.

• All the voices are robotic and tinny.

• All the scenes are too short.

And this makes sense.

Most TTS isn’t integrated with LLMs; it’s just interpreting text, so it can’t convey any emotion.

Video models are only good at ultra short scenes and are honestly still pretty bad.

Its one strength seems to be generating consistent, attractive faces in short bursts. Which makes sense: the primary use case for these things is influencer bots, impersonating real humans, for attention farms or marketing campaigns.

---

…Look. AI is just not good for this format. I doubt it ever will be without some serious cinema scaffolding to just hold bits in place.