OpenAI, recently introduced a groundbreaking tool called Sora. This new model has the ability to generate videos based on text prompts, producing realistic footage up to a minute long that aligns with user instructions regarding subject matter and style. Sora is also capable of creating videos from still images and extending existing footage with new material.
Prompt: Photorealistic closeup video of two pirate ships battling each other as they sail inside a cup of coffee.
In a blog post, OpenAI expressed their mission to teach AI to comprehend and simulate the physical world in motion. Their goal is to develop models that assist people in solving real-world problems requiring interaction. The company showcased several initial examples, including a movie trailer featuring a 30-year-old space man donning a red wool knitted motorcycle helmet, set against a blue sky and salt desert, with a cinematic style and vivid colors. These examples demonstrated the impressive capabilities of Sora.
To ensure the product’s compliance with OpenAI’s terms of service, the company invited a select group of researchers and video creators to “red team” the tool. This process involves testing the model’s susceptibility to generating content that violates OpenAI’s guidelines, which prohibit extreme violence, sexual content, hateful imagery, celebrity likeness, and the use of others’ intellectual property.
While access to Sora is currently limited to researchers, visual artists, and filmmakers, OpenAI’s CEO, Sam Altman, shared video clips on Twitter in response to user prompts. These clips were created using Sora and are marked with a watermark to indicate their AI origin.
OpenAI’s introduction of Sora follows the successful launch of their still image generator, Dall-E, in 2021, and the generative AI chatbot, ChatGPT, in November 2022, which quickly gained 100 million users. Other AI companies have also ventured into video generation, but their models have been limited to producing a few seconds of footage that often deviates from the provided prompts. Google and Meta have announced their ongoing development of generative video tools, although these tools have not yet been released to the public.
In terms of training data, OpenAI has not disclosed the exact amount of footage used to train Sora or its specific sources. However, they did mention that the training videos included publicly available content and licensed material obtained from copyright owners. OpenAI has faced copyright infringement lawsuits in the past due to their generative AI tools, which rely on large datasets scraped from the internet to imitate images and text.
The introduction of Sora represents a significant advancement in AI technology, pushing the boundaries of what is possible in video generation. OpenAI’s commitment to responsible usage and compliance with their terms of service ensures that the generated content aligns with ethical guidelines. As the AI industry continues to evolve, Sora and similar tools pave the way for exciting possibilities in various fields, including entertainment, marketing, and education.
Prompt: New York City submerged like Atlantis. Fish, whales, sea turtles and sharks swim through the streets of New York.
In conclusion, OpenAI’s Sora is a remarkable tool that can generate realistic videos based on text prompts. With its ability to adhere to user instructions on subject matter and style, Sora opens up new avenues for creative expression and problem-solving. As AI technology progresses, it is essential to strike a balance between innovation and responsible usage, ensuring that these powerful tools benefit society while respecting ethical boundaries.
All images and original post can be found here: https://openai.com/sora
