Sora: OpenAI Unveils AI Tool for Video Creation from Text Descriptions [VIDEO]
On February 15th, OpenAI, the developer of the ChatGPT chatbot and the DALL-E image generator, introduced a new artificial intelligence system called Sora, which creates realistic videos up to 60 seconds long based on user-provided text prompts. Its capabilities can be explored on the project's page.
Sora has the ability to create complex scenes with multiple characters from text descriptions, accurately depicting their movements and detailing the surrounding environment. The videos are of high quality at 1080p resolution and can compete with visual effects seen in video games and Hollywood blockbusters.
Sora adeptly renders both animals and humans, capturing their emotions through facial expressions. The tool enables the creation of cinematic sequences with various camera angles and smooth camera movements. Physical object properties, fluid simulations, and texture detailing are also implemented at a high level.
The neural network deeply understands language, allowing it to precisely interpret user requests. It comprehends not only what needs to be depicted, but also how these objects interact in the real world.
However, OpenAI acknowledges that Sora has its weaknesses. For instance, in some cases, people and animals may appear out of place and perform inconsistent actions.
To generate a video using Sora, a clear request similar to the process of working with another OpenAI product, DALL-E, which generates images from text, is required. Typically, this involves a brief description of what should happen in the video, along with instructions regarding style, colours, the number of objects and characters, and so forth.
Currently, access to the new technology is limited. Before making Sora available to the general public, a series of security measures are planned to be implemented. The company is collaborating with experts in disinformation and information security to identify potential risks. Additionally, tools are being developed to detect AI-generated content produced by Sora.
The Gaze previously reported that the United Kingdom plans to invest £100 million in creating nine new research centres in the field of artificial intelligence and training regulatory bodies to work with this technology.
Approximately £90 million will be allocated to the establishment of centres focused on using AI in areas such as healthcare, chemistry, mathematics, as well as cooperation with the United States on the responsible use of AI.