OpenAI, supported by Microsoft, is developing Sora, a video-generating software capable of creating minute-long videos from text prompts.
Currently in red teaming to uncover AI vulnerabilities, Sora also aids visual artists, designers, and filmmakers in evaluating the model, the company detailed.
The software excels in crafting intricate scenes with numerous characters, diverse motions, and detailed backgrounds, capable of producing varied shots in a single video.
Examples of prompts include historical visuals of California’s gold rush and vibrant scenes from Burano, Italy, featuring a Dalmatian and bustling canal life.
Beyond video creation from prompts, Sora can bring still images to life, as shared in a company blog post.
Following the success of ChatGPT in 2022, Sora extends OpenAI’s GenAI innovations, previously marked by capabilities in email composition, coding, and poetry. Meta Platforms, aiming to rival Microsoft, Google, and Amazon in generative AI, enhanced its Emu model for video editing and creation from text last year.
Sora remains under development, facing challenges with spatial accuracy and camera movement adherence. OpenAI is also crafting tools to identify Sora-generated videos, ensuring the discernibility of AI-created content.
