Note: these people themselves https://twitter.com/BenjaminDEKR/status/1772826040735285696 and are not AI-generated, but all their speech, facial expressions and motor skills are
What makes the above video so special is that it captures that organic ‘chaos’ of a freehand selfie video so flawlessly. Both the camera and person move. And that leads even ‘experts’ to wonder a little https://twitter.com/gurpreet395/status/1772764766278123607. ‘Creator’ Beck, who is the person of whom the AI version above is speaking, explains: “This video looks more “realistic” than the other ones out there because I worked around her “pre set” expressions. That’s real https://twitter.com/beckylitv/status/1772765788870451479. Still very much a human touch, just on the backend.“In short, it’s a matter of good ‘prompting’. But let us continue to guard against concepts such as ‘prompt engineer‘ and ‘prompt artist‘, because it remains only a matter of writing a maximum of about five sentences.
Anyway, on top of this AI development, artists have now made use of OpenAI’s video generator Sora, resulting in yet another very impressive text-to-video productions. Mainly ‘Airhead‘ comes in – see after break. Of course, OpenAI has also already begun their ground invasion of Hollywood.