OpenAI Sora launched to convert AI text into video

Sora is a new generative Artificial Intelligence (AI) model by OpenAI that creates videos from textual content. This new model is currently in preview.

It can make photorealistic videos up to 60 seconds long by understanding how things exist in the real world. It will also integrate various pictures without character or style disruption.

According to OpenAI, the new model can create highly detailed scenes with complicated camera motion and several characters. Technically, Sora is a diffusion model. It starts with a video that seems like static noise and then progressively transforms into the final result by reducing the noise step by step.

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

OpenAI highlights a problem they solved in Sora, which keeps the subject constant even when it is temporarily out of view. It also preserves the visual style by allowing the model to operate on multiple frames simultaneously. It gives the ability to predict what will happen and plan for it.

The company displayed numerous excellent movies made with Sora, including a historical video of California during the gold rush. A gold rush was a place where a large number of people moved to search for gold where it had been discovered in California. The video portrays a beautiful woman walking down a Tokyo street, a golden dog playing in the snow, and more.

Prompt: “A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. she wears a black leather jacket, a long red dress, and black boots, and carries a black purse. she wears sunglasses and red lipstick. she walks confidently and casually.… pic.twitter.com/cjIdgYFaWq
— OpenAI (@OpenAI) February 15, 2024

Click here to find out why OpenAI board fires CEO Sam Altman in a surprise move

However, OpenAI acknowledges that some of their created videos may represent unimaginable motion. For example, a guy going the wrong way on a conveyor belt or sand turning into a chair and moving in a way that does not make sense.

However, Sora is not yet available to everyone because the company is working to improve its safety. This includes rejecting text input prompts that contain severe violence, sexual content, and offensive imagery. It also rejects content that violates third-party IP or celebrity privacy rights.

OpenAI says it is collaborating with specialists in areas like misinformation, hateful content, and bias. It is to explore the limits of this new model. They intend to apply the safety measures developed for its model DALL-E-3 and C2PA metadata to test videos created through AI.

However, Sora is not the first text-to-video AI model to hit the market. Other options include Runway, Pika, Stability AI, and Google Lumiere.