Sign up for the Newsletter!
Home Artificial intelligence SORA: A new technological leap in video generation from OpenAI

SORA: A new technological leap in video generation from OpenAI

0

Sam Altman and his team once again leave us speechless after their latest presentation at SORA Artificial Intelligence

Sign up for the Newsletter!

Less than two years ago — September 2022 — I was trying to create through DallE an image of a group of people in a gym with a result that, although it could be valid to exhibit in some abstract art exhibition, It was not what is called "faithful to reality".

We have moved from images to sound and music, from search engines to the Great Models of Language —LLM— and even now Let's go for action models like the Rabbit R1 that is capable of executing tasks.

Today — well, technically yesterday — we have taken another giant leap in terms of content generation through artificial intelligence, and the OpenAI team led by the charismatic Sam Altman saw fit to introduce ourselves to SORA, a new model of video generation through AI.

An example they showed through X.com

And the leap has already been made. We have gone from DallE, Midjourney and Stable Diffusion and from generating images in our daily lives as if it were something common to starting to fool around with the idea that we can imagine any possible scenario and create it with video generation through artificial intelligence.

We have to live in a "strange" future: AI and "reality."

Last night I was watching the videos created by the community and, although we will now talk about the quality of the videos, the phrase that kept running through my head was precisely that one… We have to live a strange future.

And I think it is so because if in the space of 2 years we have gone from barely making an anthropomorphic figure - or marveling at an avocado-shaped chair - to being able to create a video in the style we want with 4K resolution and fuel to reality, What will the next two years hold for us regarding Artificial Intelligence?

I guess it's just a matter of —a little— time to find out. Seeing the pace at which OpenAI is progressing in terms of its presentations and that despite its financial difficulties, it seems that the collaborations with Microsoft are having an effect, I wouldn't be surprised if within a year this video-generating AI is at the level of the images in, for example, Midjourney.

Image generated by Midjourney in 2024. runtheprompt.com

SORA, artificial intelligence and the future of cinema.

After the presentation I could only think about the capacity that someone in the house will have. create your own movies, solely and exclusively through writing a prompt —instructions on what you want—and a computer.

This once again democratizes the art world, which I think is incredibly wonderful. It has already happened on several occasions throughout history, and it is that cheap and accessible paint empowers painters, DAWs encourage music creation and this will allow anyone to create their stories in moving images.

Can you imagine having an idea and being able to make your own series with it? I'm not saying it will be easy or quick - there's still a long way to go - but it seems like a possible future in the short rather than the long term. I'm very sure of that.

Some examples of what SORA is capable of.

Sam Altman from his Twitter account —X— was asking users to give him prompts to create with SORA, so that even though the model was not accessible to everyone, we could see the potential it had. And it was jaw-dropping…

Failures, coherence and details

There is still a way to go. We already mentioned it and despite how absurdly impressive the model itself is and considering that it is only presented without access to the public, You can see the most common failures in this type of generation through artificial intelligence.

Coherence failures, elements appearing alone or getting lost. All of this will be corrected, but until then, we can start to marvel — once again — for as long as it takes until we start to see it as something normal, as has happened in the rest of generative AIs.

Is using SORA from OpenAI possible?

Well unfortunately at the moment it is not possible to use sora individually. At the moment they have only released the project for OpenAI researchers and people who are part of the team, but not for the rest of us.

However, you can use SORA indirectly through Sam Altman and his your official account.

We are definitely living in strange times…

Exit mobile version