OpenAI, one of the leading companies in the field of artificial intelligence, announced Sora, a new video production model that can create realistic and imaginative scenes from text instructions. The OpenAI Sora text-to-video model, which can create photorealistic videos of up to one minute based on commands typed by users, combines technology and creativity.
Sora’s innovative features
Sora allows for a wide range of creativity, including complex scenes with multiple characters, specific types of movement, and accurate details of the plot and background. According to OpenAI’s promotional blog post, the model can understand how objects exist in the physical world, accurately interpret props, and produce expressive characters that express vivid emotions.
Demos created with Sora include an aerial scene of California in its heyday, a video of what appears to be a train through Tokyo, and others. However, OpenAI notes that the model may have difficulty accurately performing a physical simulation of a complex scene.
Recently, companies such as Runway and Pika have attracted attention with their own text-to-video models, while Google’s Lumiere stands out as one of OpenAI’s main competitors in this field. Similar to Sora, Lumiere offers users tools to create video from text and also allows creating video from a still image.
For now, OpenAI Sora is available to “red team” members who evaluate the model’s potential harms and risks. On the other hand, OpenAI also provides access to some visual artists, designers, and filmmakers to get feedback. The company notes that the current model may not accurately reproduce the physical laws of a complex scene and may not accurately interpret certain cause-effect patterns.
OpenAI, like other AI products, will have to deal with the consequences of AI-generated videos being mistaken for real. In this regard, the company stated earlier this month that it had added watermarks to its text-to-image tool DALL-E 3, but that they could be “easily removed.” Sora brings about important discussions not only in the world of technology and art, but also at the ethical and social level.