Hollywood producer Taylor Perry, who recently slashed a $800 million studio expansion, revealed, "With Sora, we don't have to travel abroad to shoot." As a veteran filmmaker who writes, directs, acts, and produces, Taylor Perry is clearly happy to see project costs cut dramatically because of AI.
With the rapid growth of generative artificial intelligence (AIGC) in the entertainment industry, Hollywood is evaluating the potential of this technology to be incorporated into the film and television production process.
Sora, a text-generated video model launched by artificial intelligence research company OpenAI in February this year, has been a hot topic in the field recently. Sora is able to create realistic video clips based on text prompts of a few sentences. OpenAI has released a series of test videos on its official website, looking forward to an ongoing dialogue with artists and creatives about the Sora application, and hopes to work with the industry to roll out the latest technology in phases during the iterative deployment process. At the same time, participants in the Hollywood film and television industry have had a lot of discussions about the changes brought about by generative AI represented by Sora and the future development space.
Shy Kids, a Toronto-based multimedia production company, was one of the first teams to try Sora. Founded by Walter Woodman, Sidney Leader and Patrick Seidberg, the company has worked on several film and television projects with HBO, Disney and Netflix. The trio used Sora to create a surrealist short film called Air Head, which features a man with a balloon head.
Patrick Seidberg revealed that Sora's process of generating the image of the protagonist of "Balloon Man" is not as simple as imagined, "When we type in 'a man with a balloon head', the AI will generate an image based on this prompt, but most of it is not very creative, such as a small smiley face drawn with a marker. It takes a lot of trial and error to tweak the prompts until the system makes something that meets the team's requirements. ”
Balloon Man
Walter Woodman adds, "We tried at least a hundred times. The discussion on the Internet makes it feel like 'it's easy to generate AI internals', but in fact, if you look at the failed content on our computers, you will understand that it is not an overnight thing, and it requires a lot of trial, tweaking, and correction. A lot of times, a shot has a perfect composition, but the color is wrong, and we have to process and recolor the image. There were also times when the system was never able to generate the content we needed, and we had to manually erase the excess and manually adjust the VFX to achieve the effect we wanted. ”
Walter Woodman concludes that Sora can be seen as a tool that simply says, "Tools are meant to help humans create, and a lot of people have been talking about how powerful generative AI is lately, as if it's going to replace everything." In my opinion, this statement seriously devalues the creativity required for a story, the elements that make up a story, and the importance of the storyteller. ”
Of course, Sora's potential is undeniable. "What excites me most about AI is being able to bring the idea in my head faster and present it better," said Patrick Seidberg. At the same time, I hope that AI can break down the barriers and barriers that currently exist in the film and television industry. In a word, Sora has the ability to democratize the film industry. ”
For example, the widespread use of generative AI in the filmmaking process depends largely on how U.S. federal courts rule on copyright issues raised by this emerging technology. The biggest question for AI-generated content today remains on copyright: whether the use of copyrighted material to train AI models constitutes infringement, and whether AI-generated works are eligible for copyright protection.
It's worth noting that OpenAI has never disclosed the source of the material used to train its system, and the reason given by the company is that the reason for not disclosing the source of the data is to maintain a competitive advantage over other companies. However, the company has received lawsuits from several well-known writers and writers alleging that OpenAI used their copyrighted works for system training, most of which were downloaded from pirated websites.
For the Hollywood film and television industry, 2024 is an important stage for AI to verify the logic of removing the false and preserving the true. Investors and the market will test whether the application of AI technology can achieve efficient output, improve productivity and meet market demand in more complex video, gaming and other fields. Once these film and television entertainment companies with mature commercial value prove that they can rely on AI to reduce costs and increase efficiency, it means that AI will complete a milestone breakthrough in penetration from scratch.
Special Planning|AI Pathfinder Film and Television New Quality Productivity
-
Radio and television accelerate the layout of AIGC and focus on the development of new quality productivity
- Technology empowers the application of artificial intelligence in film and television
- The online audio-visual platform "grabs" artificial intelligence
- AI Creation: Broaden Your Horizons and Hold Your Boundaries
- Interim Measures for the Administration of Generative AI Services
- Sora's limitations and potential
- AI helps creators break through barriers
- The world's first Artificial Intelligence Act
Submissions should be sent to: