While I still have deep convictions regarding Phantom and its quality of earnings, especially after the recent performance numbers, I can’t help but fathom the absolute carnage that is bound to come their way due to text-to-clip innovations. Some of those disruptors are among the most hyped startups of our age (see https://openai.com/sora).
While I stand by my opinions on Generative AI being a rather distant threat, the quality of the outputs presented by OpenAI (especially with respect to minute-long videos) are both impressive and potentially threatening to companies like Phantom and Basilic. To say the least, they have shown that merely text-to-video models are enough to generate a 95%-studio-quality clip. Assuming you could finetune these models by incorporating existing camera shots into the prompts (which I’m sure people are already working on), you could potentially get a 99%-studio-quality clip. At that point, most film-makers would simply switch to these lightning fast and cheap generative models, instead of having to wait for an expensive human designer to do the touch-up.
Given that an average, uninterrupted movie shot is about a minute long, I think a lot of the touch up work on movie sequences at the very least is going away from VFX studios towards GenAI startups. Obviously, a full-on movie contract would still be handled by large studios, but there are certainly signs of potential damage here.
Disc: Recently sold out (position size in previous post).
Subscribe To Our Free Newsletter |