Mar 25, 2024 20:00:00
OpenAI, which develops the large-scale language model GPT and the image generation AI model DALL-E , announced ‘ Sora ‘, an AI that generates movies from text, in February 2024. Regarding Sora, Bloomberg reports that OpenAI is meeting with Hollywood movie studios, media executives, and talent agencies to promote Sora for use in movie production.
Bloomberg – Are you a robot?
https://www.bloomberg.com/news/articles/2024-03-22/openai-courts-hollywood-in-meetings-with-film-studios-directors
Sora is an AI that can generate up to one-minute movies from text, simulating the laws of physics to create complex scenes with multiple characters, specific motions, and precise details of subjects and backgrounds. It has the characteristic that it can be generated while understanding whether it exists. There have been AIs in the past that generate movies from text (prompts), but past AIs were unable to understand the cause and effect of movements, so for example, “I made a movement to bite a cookie, but the cookie ate me.” Movies containing contradictions such as ‘No trace left’ were often generated. However, because Sora uses physics simulation to ‘understand the real world,’ it can generate movies that are more natural and realistic.
OpenAI announces ‘Sora’, an AI that generates videos while simulating physical laws from text – GIGAZINE
At the time of Sora’s announcement, OpenAI released several demo movies. For example: ‘A stylish woman with red lipstick in a black leather jacket, long red dress, black boots, black purse and sunglasses walks down a Tokyo street with neon and animated signs. She looks confident. A demo movie generated with Sora from the text ‘Many pedestrians are walking around the city.’
Sora is capable of scaling the world through physics simulation, and is expected to be able to do a variety of things, such as dynamic camera motion based on a consistent 3D image and the ability to fuse two completely different movies. In addition to physics simulation, it is also possible to reproduce games, illustration-based animations, and movies that cannot exist in reality. Below is a movie of “forming “SORA” with realistic clouds” shown in the Sora demo.
At the time of writing, OpenAI has not announced when it will release Sora. However, according to people familiar with the entertainment industry, OpenAI has been pitching Sora to Hollywood even before its release.
OpenAI Chief Operating Officer Brad Lightcap will hold an initial meeting with Hollywood film executives in late February 2024, anonymous sources told Bloomberg. He said he did. Mr. Lightcap and several colleagues demonstrated the process of generating movies with Sora. A few days later, it was confirmed that OpenAI CEO Sam Altman attended a party in Los Angeles after the Academy Awards ceremony.
According to Bloomberg, Sora has not been released to the public as of March 2024, but OpenAI has distributed access to Sora to some famous actors and directors. In announcing Sora, an OpenAI spokesperson said, ‘OpenAI uses an iterative deployment process to deploy AI advances in stages to ensure the safety of our AI and give people ideas for future deployments. ‘Throughout this process, we have had a deliberate strategy to collaborate with the entertainment industry. We look forward to continued dialogue with artists and creators.’
Opinions about AI are divided in Hollywood, leading to strikes demanding that the use of AI that takes away the jobs of screenwriters and actors be regulated. In response, OpenAI stated that it would take safety measures, saying, “We will work with a red team made up of experts in fields such as misinformation, hateful content, and bias to adversarially test our models.” , explicitly states that the metadata will include an element indicating whether the movie was generated by Sora.
Hollywood actors and screenwriters strike for “regulation of AI that takes away jobs” – GIGAZINE
OpenAI remains at the forefront of AI development, but there have beenreports that Apple may adopt Google’s AI, Gemini , for the iPhone, and Google has announced that it is possible to use Google’s AI, Gemini, for the iPhone. They have announced the AI “ VLOGGER ,” which generates “realistic videos of people talking with body language,” and many AI startups are rapidly growing. ,” Bloomberg noted.
