The company behind the ChatGPT model, OpenAi, introduced Sora. It is a tool for creating videos with artificial intelligence with a mechanism as simple as that of its textual chatbot and similar to what we find in tools like Midjourney or DALL-E 3.
It is essentially enter instructions via text (prompt) and get a unique AI-generated video. A real breakthrough which, even if it still requires improvement, impresses with its level of realism and lays important foundations for amateur and professional creations in the audiovisual industry.
From creating realistic images with AI to creating videos: that’s Sora
The explosion of generative artificial intelligence over the past year and a half, led precisely by OpenAI, is impressive. More or less correctly, the generation of static images based on textual instructions was already setting a good part of the trend. Exactly, ChatGPT Plus integrates DALL-E 3 for this.
What we want now with Sora is to go further and generate artificial videos up to 60 seconds with a hyper realistic style. As long as the examples that OpenAI has already shown They could easily pass for a recorded and edited video, even though the company claims to have done no editing.
Here, the magic will lie in the user’s ability to generate a precise prompt. It’s exactly the same with images. We will be able to tell the action we want, the scene, the type of light, the characters and their descriptions and everything that comes to mind.
In addition to generating videos from scratch, Sora will also be able to edit existing videos, enlarge them, and even create a video animation of a still photograph.
But if there is anything to emphasize, it is that It has other capabilities like being able to extend an existing video or modify it. For example, a video of a person walking in the city turns it into a person walking in the middle of the jungle. Or merge two videos into one. And all this with the possibility of generating them horizontally, vertically or in 1:1 square format.
This, yes, It should also be noted that this is not a perfect model.. Not yet. They explain in OpenAI that the system may not represent a complex scene accurately enough and may not understand cause and effect relationships. However, generally speaking, it is also said that Sora is also able to understand what the world is like..
We can try Sora on iPhone, iPad and Mac (but not yet)
The million dollar question of the last few hours is how to access Sora. Well, it will be another OpenAI product as the company itself has confirmed. However For the moment it is reserved for laboratory tests and is not accessible to the public.
They say at OpenAI that they are still working on this model to not only refine it as much as possible, but they also want to make sure it is safe for users. They have not offered dates when it will be available.but we hope that it will not take too long and that it will eventually arrive in the coming months.
It remains to be seen whether she will be paid or not.. From what has been announced now, Sora is capable of creating videos of up to a minute in length, so this feature could well be something accessible to everyone for free and that of a longer duration would involve subscribing with a recurring payment. In any case, these are pure hypotheses, since nothing has been announced in this regard either.
More information | OpenAI
In Applesfera | iPhone 16 leak gives definitive clue about the future AI Apple will integrate into iOS 18
In Applesfera | Meet Pi, the new artificial intelligence that triumphs in WhatsApp iOS
