What is Sora?
OpenAI, the maker of ChatGPT, has launched a new AI model: Sora. Sora can create realistic and imaginative videos from text instructions. It is a diffusion model that starts with noise and makes it into a video scene that matches the user’s text. Sora can make videos up to a minute long with good visual quality and relevance to the user’s text. Sora can also make videos from a still image or add new material to existing footage.
How does Sora work?
Sora belongs to a class of AI models that can turn written descriptions into video content. This is a hard task that needs to understand the moving physical world, simulate camera movements, and make multiple characters with emotions. Sora uses a transformer architecture like GPT, which makes it scalable and efficient. Sora learned from a large collection of videos that were public or had permission.
What can Sora do?
Sora has many possible uses in different areas, such as education, entertainment, art, and journalism. For example, Sora can make educational videos from text summaries, movie trailers from scripts, artistic videos from creative texts, or news videos from articles. Sora can also help users see their ideas and stories in a more interesting way.
How can I access?
OpenAI has given access to Sora to some researchers and video makers who will test the model for what it can and cannot do. The company has also shown some examples of videos made by Sora on its website and social media. The videos have a watermark to show they were made by AI. OpenAI has also said that it will watch how Sora is used and follow its rules, which do not allow making videos of extreme violence, sexual content, hateful images, celebrity faces, or the IP of others.
Sources:
I really enjoyed your unique approach to this topic. Your article was both informative and innovative.
Thank you