|
Listen to this story
|
OpenAI is now living the Hollywood dream with its text-to-video platform Sora. AIM won’t be surprised if Sora helps artists and filmmakers elevate their creativity to the ultimate levels and someday win the Oscars and Walk of Fame.
It’s been a busy few months for OpenAI, and there are still no signs of GPT-5. Meanwhile, the hottest AI startup is arranging meetings in Los Angeles with Hollywood studios, media executives, and talent agencies, trying to forge partnerships in the entertainment industry and encourage filmmakers to integrate their latest video generator, Sora, into their work.
It even published a blog illustrating how artists, filmmakers, and creative designers can use Sora to generate surreal videos, which can change the way Hollywood, advertising, and other creative industries work.
In a recent interview, OpenAI chief technology officer Mira Murati said that Sora would be available this year, possibly within “a few months”. However, she appeared hesitant about delving into the specifics of the data Sora was trained on and dodged the question.
“I won’t go into the details of the data used, but it was either publicly available or licensed data,” she said. Murati was trolled for saying that she wasn’t sure whether it used videos from YouTube, Facebook, and Instagram. However, she confirmed that Sora uses content from Shutterstock, with which OpenAI has a partnership.
Comes With a Price
Murati said that Sora could take a few minutes to generate videos, depending on the complexity of the prompt, and that it’s ‘much, much more expensive’. “We don’t know what it’s going to look like exactly when we make it available to the public, but we’re trying to make it available at a similar cost eventually to what we saw with DALL-E,” she added.
The average production cost for a major Hollywood studio movie is approximately $65 million, while the cost of one NVIDIA H100 is $30,000. Training Sora demands substantial compute power, with an estimated requirement ranging from 4,200 to 10,500 NVIDIA H100 GPUs for a month-long training period, according to a report by Factorial Funds.
The inference cost for diffusion-based models like Sora is significantly higher than that of LLMs, reaching multiple orders of magnitude. If Sora becomes publicly available, approximately 720,000 NVIDIA H100s GPUs would be required to support the TikTok and YouTube communities alone.
The subscription cost for Sora is expected to be high, leading to the likelihood that OpenAI may keep the access limited to filmmakers, studios, and artists. In a poll on X, a majority of the respondents declined to pay $99 per month for a Sora subscription.
If OpenAI plans to make Sora available to the general public, the computational demand will increase significantly. Therefore, hosting Sora at the price of DALL-E 3 is not feasible.
Perfect Sora Doesn’t Exist
A Toronto-based multimedia production company, Shy Kids, created a short film titled Air Head using Sora, featuring a protagonist with a balloon for a head.
While the short film garnered quick praise, it is crucial to acknowledge that AI-generated films are yet to achieve consistent character portrayals. “True character consistency with AI is quite challenging. You can fly a drone through the tunnel, but to maintain character consistency, you have to use a yellow balloon,” wrote a user on X.
AI Researcher at Google, Jon Barron, wrote, “Generative video technology (e.g. Sora) has two huge challenges in front of it that will likely slow adoption by Hollywood et al: 1) high latency (waiting minutes/hours to get seconds of footage), and 2) lack of controllability (the video you get back isn’t necessarily what you wanted). These are hard problems to solve, but history suggests we only need to solve one of them.”
Only time will tell how serious OpenAI is about Sora as a creative tool for filmmakers. “OpenAI is an AI company, not a VFX company. Besides not even understanding the needs of their users, they see this model as a neat intermediate result on their path to the AGI, and as a progress report to raise more money,” wrote a user on Hacker News.
“They’re really interested in advanced emergent behaviour it exhibits, not in artistic tools. For this reason, they’ve never bothered to fix all the artifacts DALL-E 3 gives, let alone add any tooling to it. Sora will be the same, and its quality doesn’t even remotely approach what is required in production. It’s more of an experiment,” the user added.
Despite the limitations, there’s no denying that Sora could be a key tool for OpenAI to generate massive revenues. It would also be interesting if the company released a movie about Sam Altman’s ousting and the mysterious disappearance of Ilya Sutskever using Sora.
All in all, OpenAI is nothing without drama, and what better place for that than Hollywood?
