Leader in artificial intelligence OpenAI unveiled a new model dubbed Sora, which it says can produce 60-second videos that are both “imaginative” and “realistic” with short text inputs.
The business said on Wednesday that Sora can create films up to 60 seconds in length based on text directions. It can provide scenes with several actors, distinct animation styles, and intricate backdrop features.
According to the blog post, “The model comprehends not only what the user has requested in the prompt, but also how those things exist in the physical world.”
As stated by OpenAI, the goal of training the AI models is to enable them to “assist people in solving problems that require real-world interaction.”
This is the most recent endeavour from the business that created the well-known chatbotChatGPT, which keeps advancing the field of generative AI. Reece Hayden, a senior analyst at market research firm ABI Research, believes that OpenAI’s Sora stands apart due to its length and accuracy, even if “multi-modal models” are not novel and text-to-video models already exist.
New personalised material being distributed across channels by these kinds of AI models, according to Hayden, may have a significant influence on the digital entertainment industries.
According to Hayden, “one obvious use case is within TV; creating short scenes to support narratives.” “The model indicates the direction of the market, but it is still limited.”
However, OpenAI noted that Sora is still a work in progress and has obvious “weaknesses,” especially when it comes to cause and effect and spatial elements of a prompt, such as mixing up left and right. The example of making a film showing someone biting into a cookie without leaving a bite mark was provided.
OpenAI’s messaging is still centred on safety for the time being. The business stated that in addition to testing the most recent model, it would thoroughly examine a number of topics, such as prejudice, hate speech, and disinformation, in collaboration with a group of specialists. According to the company, it is also developing technologies to assist in identifying false information.
Professors of cybersecurity, or “red teamers,” will be the first to have access to Sora so they can evaluate the product for dangers or drawbacks. In order to get input on how creative professions may utilise it, a handful of visual artists, designers, and filmmakers are also being granted access.
The most recent information is provided as OpenAI works to improve ChatGPT.
The business said earlier this week that ChatGPT is testing a feature that lets users instruct the platform to remember prior discussions in order to personalise future interactions, or to forget what was talked in the past.