Artificial intelligence research and deployment company OpenAI introduces their new Video Generation Model, Sora. OpenAI boasts of its captivating and high-quality video quality. It is a diffusion model and uses previous research data from DALL-E and GPT models. This text-to-video generation AI model can generate realistic videos from text prompts, images, or other videos. This model is still in development, and we might see it implemented in GPT-4 or GPT-5 in the future.
OpenAI showcases videos made with Sora
OpenAI published several videos generated with Sora. These video clips range from 8 seconds to 1 minute. Sora shows excellent continuation over several shots in videos. It follows characters in a detailed environment and can simulate real interactions in the physical world. It can generate videos from a still image and extend a previous video. Also, it can fill in the missing frames of a video. The vibrant, detailed, and photorealistic videos generated with Sora are captivating.
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/7j2JN27M3W
Prompt: “Beautiful, snowy… pic.twitter.com/ruTEWn87vf
— OpenAI (@OpenAI) February 15, 2024
It can simulate character movements, lights, dust, and camera movements. OpenAI says that the AI model knows how things look and interact in the real world. This knowledge allows it to simulate real-life scenarios vividly. There are some limitations to the current video generative model. Researchers found out that Sora might have some difficulty simulating complex physics. It may be difficult for the AI to recognize cause and effect in a scene, such as bite marks not appearing after biting a cookie.
Safety and ethical usage
The research was led by Bill Peebles & Tim Brooks. This AI model is currently available for Red Teamers to check critical areas for risks. OpenAI also granted access to the model to some visual artists, filmmakers, and designers for feedback. Before being released as an OpenAI product, Sora will need to pass intensive safety tests that ensure AI’s safe and ethical usage.
OpenAI plans to implement C2PA Metadata so that videos made with Sora can be tracked to their origin. OpenAI uses this feature in DALL-E 3 to reduce AI abuse. Sora will also use a text classifier to block illegal or abusive prompts. Sora isn’t available to the public yet. But it shows what the future holds and gives us some ideas of what AGI will be capable of.
YouTube: OpenAI Sora – The Age Of AI Is Here
By clicking play, you agree to YouTube's Terms of Service and Privacy Policy. Data may be shared with YouTube/Google.
Photo credit: The feature image is symbolic and has been done by Christopher Isak with Midjourney for TechAcute.
