Everything you need to know about SORA

New revolutionary text to video AI model from OpenAI

Odair Faléco
4 min readFeb 17, 2024

Imagine a world where you can conjure up any visual reality you can dream of, from soaring through the Grand Canyon on the back of a dragon to exploring the rings of Saturn up close, and make from it a 1 minute video. That’s the potential of Sora, the latest mind-bending creation from OpenAI.

Think of Sora as a Hollywood special effects artist on steroids. But instead of pixels and polygons, it wields the power of artificial intelligence to paint moving pictures onto the canvas of your imagination. It’s like giving a toddler a paintbrush and a vat of imagination, except instead of scribbles on the wall, you get hyper-realistic videos that could fool even the most discerning eye. The quality is impressive, and it’s a leap forward, making current AI video tools look obsolete.

But Sora isn’t just about pretty moving pictures. It’s about understanding the world in a way that computers never have before. By training on a massive dataset of videos, Sora has learned the intricate dance of cause and effect that plays out in our physical world. It can predict how objects will move, how light will interact with them, and even how things will sound.

This ability to understand the world’s physics makes Sora a powerful tool for a variety of applications. Imagine using it to train robots, design safer cars, more efficient airplanes, or even new medical treatments. The possibilities are truly endless.

See video samples from the SORA AI video model

As computers become more powerful, they are starting to blur the lines between the real and the virtual. Video generation models are a prime example of this trend. These models can create worlds that are so realistic that they can fool our senses. This raises some interesting questions about the nature of reality and our place in it.

Sora is a diffusion model, given input noisy patches (and conditioning information like text prompts), it’s trained to predict the original “clean” patches. Importantly, Sora is a diffusion transformer. Transformers have demonstrated remarkable scaling properties across a variety of domains, including language modeling, computer vision, and image generation.

Here are some technical details:

  • Sora is a large language model, trained on a massive dataset of text and code.
  • The model can generate videos of different lengths, resolutions and aspect ratios.
  • It can also be prompted with text or images to create videos.
  • It can also extend videos (from beginning or end) and create video loops
  • It can make multple scenes of the same video with consistency
  • The researchers believe that this is a promising approach for developing general purpose simulators of the physical world.
  • 3D consistency — Sora can generate videos with dynamic camera motion. As the camera shifts and rotates, people and scene elements move consistently through three-dimensional space.

Long-range coherence and object permanence — A significant challenge for video generation systems has been maintaining temporal consistency when sampling long videos. We find that Sora is often, though not always, able to effectively model both short- and long-range dependencies. For example, our model can persist people, animals and objects even when they are occluded or leave the frame. Likewise, it can generate multiple shots of the same character in a single sample, maintaining their appearance throughout the video.

Interacting with the world — Sora can sometimes simulate actions that affect the state of the world in simple ways. For example, a painter can leave new strokes along a canvas that persist over time, or a man can eat a burger and leave bite marks.

Simulating digital worlds — Sora is also able to simulate artificial processes–one example is video games. Sora can simultaneously control the player in Minecraft with a basic policy while also rendering the world and its dynamics in high fidelity. These capabilities can be elicited zero-shot by prompting Sora with captions mentioning “Minecraft.”

Of course, there are also some potential downsides to consider. With great power comes great responsibility, as Uncle Ben wisely said. In the wrong hands, Sora could be used to create deepfakes that could wreak havoc on our trust in reality. But as with any powerful technology, it’s up to us to use it responsibly and ethically.

So, the next time you’re watching a movie or playing a video game, take a moment to appreciate the incredible technology that went into creating it. And then, look to the future, where AI like Sora is poised to take us on a journey beyond the limits of our imagination.



Odair Faléco

AI Artist & Consultant • UX Lead at Accenture Song, previously R/GA • Founder https://zero1cine.com • Author • Speaker • Musician | https://ai.ofaleco.xyz/