Let’s explain to you what Sora is and what he can do, the new artificial intelligence system created by OpenAI. This is a new artificial intelligence revolution that comes from the creators of ChatGPT and DALL-E, who are once again innovating in the sector.
Just as ChatGPT generates text and DALL-E generates images from text, What Sora does is generate videos from our textual commands. At the moment, this model is not accessible to everyone, only to company researchers, but it promises to be the next big revolution.
What is Sora and how does it work?
Sora is an artificial intelligence system created by OpenAI, the same company that has created other systems known as ChatGPT or DALL-E. In this case, it is an AI that generates videos from textso you write what you want to see using a prompt or textual command, and it will generate it out of thin air.
This artificial intelligence is very similar to models that create images from text, and at the same time is based on some of the technology of language models such as GPT. This system is capable of understand what you ask with natural languagesince it has been trained to understand the way we usually speak and how we express ourselves to request or ask something.
After understanding what you’ve asked at a prompt, Sora will be able to generate videos that show what you’ve asked in motion, which is a huge step forward from simply creating images.
It has achieved this because it is a trained model with a huge library of videos, so it knows how to recognize movements, descriptions and anything you ask of it, and will be able to recreate them on video. He will know what you mean when you talk to him about types of people, clothing, accessories or visual effects.
What you can do with Sora
Sora is capable of generating complex scenes in which several characters appear, and also you can specify the type of movement what you want me to do. You can also detail the environment, and Sora will know how to interpret what you say and what things are like in the physical world.
Look for example at the video that we are going to show you below. In it, a prompt has been used in which you have been told to create an elegant woman walking down a Tokyo street full of neon, and with urban signs. In the prompt that was used, her clothing was also described in detail, as well as her sunglasses or lipstick.
Therefore, it is able to accurately generate the person and reflect how you have asked them to be dressed. In addition, it also reflects other details that have been requested, such as the street being wet and having mirror-effect reflections. Below we leave you the video, and below it the translation of the prompt used.
“An elegant woman walks down a Tokyo street filled with warm bright neon and lively urban signage. She is wearing a black leather jacket, a long red dress, black boots, and a black bag. She is wearing sunglasses and red lipstick. She walks confidently and carefree. The street is wet and reflective, which creates a mirror effect of the colored lights. Many pedestrians walk by.”
This can serve as an example for you to see the qualities and reliability when it comes to reflecting what you have asked. He also understands contexts, and being in Tokyo has made the woman have oriental features. In addition to all this, Sora also allows you to specify traits like her age or her clothing, or some other physical attributes and the movements she is performing.
As usual, Sora’s abilities are also They will depend on each person’s ability to create prompts. complexes in which to make the best descriptions with which to obtain customized results. The more detailed the prompt description, the more detailed the image displayed will be regarding what you want to obtain.
At the moment, this artificial intelligence model is in an early phase of its development, and generate videos up to 60 seconds. In addition, the company warns that it may have some problems when it comes to accurately recreating some physics.
When will Sora arrive?
As we have said, at the moment this model is in a testing phase, and can only be accessed by OpenAI researchers. Therefore, it is still limited to laboratory tests, with which its operation is being tested.
Here, OpenAI ensures that Sora will end up joining the product catalog of the company, but has not yet given any specific date for this. For now, they say that they must first take measures to guarantee the safety of users.
Xataka Basics | ChatGPT detector: 9 services and apps to know if a text has been generated by OpenAI’s AI