A video of Brad Pitt and Tom Cruise is making rounds online, and it's wow.
The entire clip is 100% generated by a neural network — the new Chinese model Seedance 2.0. And it's the result of a short one-line text prompt.
What impressed me most:
Character consistency. Faces don't "drift" from frame to frame, lighting and shadows are perfect.
Emotions and lip sync. The model pulled the right tone, facial expressions, and perfectly synchronized lips with voice.
Production value. Looks like a scene from an expensive Hollywood movie, though it's just pixels assembled on the fly.
A year ago this quality required a full VFX team, a budget, and weeks of rendering. Now — a few words in a chatbot.
Video on the internet can no longer serve as any proof of reality. Video used to be fact; now it's fiction at any scale.
<video src="/media/posts/ai-kills-actors-video.mp4" controls></video>