- Hollywood Tech Nerds
- Posts
- OpenAI's Text-to-Video Debut
OpenAI's Text-to-Video Debut
Plus: The Power of Low Budget Practical Effects
Happy post-Presidents’ Day Hollywood tech nerds!
In this week’s post:
Subscribe to get Hollywood Tech Nerds magically delivered to your inbox every Tuesday!
OpenAI's Text-to-Video Debut
No two ways about it, one of the biggest pieces of tech news this past week was OpenAI’s Sora, described on their website as “…our text-to-video model [that] can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt.”
One thing that may set Sora apart is its ability to interpret long prompts -- including one example that clocked in at 135 words. The sample video OpenAI shared on Thursday demonstrate Sora can create a variety of characters and scenes, from people and animals and fluffy monsters to cityscapes, landscapes, zen gardens and even New York City submerged underwater.
Given the extensive processing power requirements for AI and its related environmental impacts, perhaps “NYC submerged underwater” is not the greatest imagery to evoke!
CNET further describes some of the applicability of the tech, “The model also can generate video from still images and extend existing videos or fill in missing frames.” This is exciting stuff for video professionals, an easy way to correct issues with footage and render some visual effects with ease.
Unfortunately, with the arrival of a product like Sora, the VC AI grifting class has also logged on, doing their best to relentlessly overhype unrealistic applications. For example:
Movie watching experience:
2005: go to a movie theater
2015: stream Netflix
2025: ask LLM + text-to-video to create a new season of Narcos to watch tonight, but have it take place in Syria with Brad Pitt, Mr Beast and Travis Kelce in the leading roles
— Matt Turck (@mattturck)
11:19 PM • Feb 15, 2024
First of all, “a new season of Narcos in Syria with Brad Pitt, Mr. Beast, and Travis Kelce” is exactly the level of creativity I would expect from a venture capitalist.
Secondly, this is the kind of overhype one must be careful of with AI: it’s not magic! If you keep promising magic results, consumers are eventually going to stop listening to you. Well, maybe not all consumers; Canada’s most gullible public intellectual Jordan Peterson was fooled into thinking a scene from the Bollywood film Singham was created by AI.
Jordan Peterson assumed the scene from Singham as an AI prompt 🤣🤣
— Shilpak. (@ugach_kahitarii)
4:50 AM • Feb 17, 2024
With apologies to Dr. Peterson, AI is not ever going to make something like Singham or Mad Max: Fury Road on its own, but it is the type of technology that would be incredibly useful when making such a movie.
The Power of Low Budget Practical Effects
On the related topic of practical effects, I loved this video from filmmaker Markus Rothkranz on his ultra low budget filmmaking methods in the 1980s and 1990s, from miniatures to faking a jet cockpit in his living room to using light stands as guns.
I love seeing this stuff, it’s a reminder that even the jankiest setup for a practical effect can often bear better results that the most intricate CGI (or AI).
Kernels (3 links worth making popcorn for)
Here’s a round-up of cool links about Hollywood and technology:
Martin Scorsese on the future of cinema. (link)
Dune: Part Two - filmed with IMAX or filmed for IMAX? (link)
How to keep your Twitter account from getting hacked like the SEC. (link)