OpenAI is currently working on an upgrade to its video AI Sora, which made its debut earlier this year in February.
According to The Information, citing sources close to the company's management team, OpenAI is developing an upgraded version of its video AI Sora. The new version aims to produce longer and higher quality video clips faster than the original demonstration.
It's worth noting that Sora wasn't fully ready at the time of its release. Initial video generation reportedly took over ten minutes, and feedback from some filmmakers was quite negative. For instance, producer Patrick Cederberg had to generate hundreds of videos to obtain a usable clip, but the model struggled with maintaining consistency in style, objects, and characters. Additionally, the system encountered some physical errors, which are not uncommon in other video models.
To enhance Sora's performance, OpenAI is gathering millions of hours of high-resolution video footage, covering various styles and themes, to reduce biases and improve the model's stability. The goal is to make the generated videos more consistent in style and content, and better meet user needs.
During this period, OpenAI also stated that Sora is still in the research phase as they are in discussions with policymakers. OpenAI's Chief Technology Officer, Mira Murati, mentioned in mid-March that Sora will definitely be released this year, but the cost of generating videos needs to be reduced, as the current system is significantly more expensive than existing AI systems.
Since Sora's launch in February, the competition in the video AI market has become increasingly fierce. In China alone, four new video AI systems have emerged, including Kelin, Vidu, Jimeng, and MiniMax's Conch Video. Run ML has also introduced two new models and has partnered with Lionsgate.
These developments indicate that the video AI field is rapidly evolving, and OpenAI's Sora must quicken its pace to remain competitive.
Key Points:
🌟 OpenAI is upgrading video AI Sora with the goal of generating longer and higher quality video clips.
⏱️ The initial version had long generation times and issues with style and physical errors.
📈 The video AI market is rapidly developing, and Sora needs to accelerate to keep up with the competition.