By Enric Llonch, on 22 August 2024
Sora is an AI tool developed by OpenAI, the same company behind ChatGPT, for creating videos from text.
As of the publication of this article it is not yet available to the general public, as they're still refining its results and ensuring its safety to prevent the generation of videos with hateful content or that convey prejudices or misinformation. Likewise, they want to ensure it does not accept inputs related to violent, sexual content, or images of other people.
However, it's only a matter of time before Sora is released. So if you want to know what it's capable of and be prepared for when it becomes available, keep reading as we reveal all the details of this incredible tool.
How Does Sora Work?
This AI tool for creating videos, developed by OpenAI, can generate up to one minute of content from text provided by users, maintaining both the indicated style and theme.
Additionally, it can generate a video from a still image, animating its content with great precision and attention to detail. Sora can even extend the duration of an existing video by adding more frames.
OpenAI based the design of this AI on previous research conducted for the development of the models behind DALL-E and GPT.
The example video included in this article was generated by the tool down to the last detail, without any post-production adjustments by OpenAI team members, according to the company's official AI website.
First Ad Created with Sora
The children's toy brand Toys"R"Us was the first to use Sora to create a complete brand ad. They did this in collaboration with the California-based creative agency, Native Foreign.
Source: toysrus.com
Above, you can see the full spot where the company wanted to convey the dreams of its founder, Charles Lazarus, from his childhood, and represent its iconic 1930s mascot, Geoffrey the Giraffe.
Although Toys"R"Us is the first corporation to create a video with Sora and release it, Sam Altman, CEO of OpenAI, has been engaging the public by responding to X users' requests with AI-generated videos.
https://t.co/uCuhUPv51N pic.twitter.com/nej4TIwgaP
— Sam Altman (@sama) February 15, 2024
Source: x.com/sama
What Can Brands Do with Sora?
AI video generation tools like Sora offer brands opportunities such as:
- Creating video content more economically and quickly
- Making possible ideas that were unachievable due to budget or professional limitations
- Enhancing the creativity of a video marketing strategy
- Utilizing existing text and image content to convert it into video or improving and extending the duration of already created videos
Sora, Still in Development
The company behind Sora indicates on the official AI website that the tool still has some weaknesses and errors. For instance, it cannot yet generate audio to accompany the video.
To gradually refine it, they are collaborating with artists, filmmakers, designers, and scientists who are testing the tool and providing feedback to help improve its results without violating safety standards.
Some of the errors made by the AI include recreating collisions between objects, representing interactions between people and objects, or understanding the direction elements are moving in. Below, you can see examples of AI-generated videos that exhibit some curious flaws.
Source: openai.com/sora
Despite the errors, Sora is the only known tool capable of generating videos with this quality. Other companies have developed similar tools, but they are not as advanced as OpenAI's, and their videos are shorter. So much so that experts like Jim Fan, a research scientist at NVIDIA (a world leader in AI computing), say that Sora aims to simulate real-world physics and express fascination with this model.
Apparently some folks don't get "data-driven physics engine", so let me clarify. Sora is an end-to-end, diffusion transformer model. It inputs text/image and outputs video pixels directly. Sora learns a physics engine implicitly in the neural parameters by gradient descent…
— Jim Fan (@DrJimFan) February 16, 2024
Source: x.com/DrJimFam
I see some vocal objections: "Sora is not learning physics, it's just manipulating pixels in 2D".
— Jim Fan (@DrJimFan) February 16, 2024
I respectfully disagree with this reductionist view. It's similar to saying "GPT-4 doesn't learn coding, it's just sampling strings". Well, what transformers do is just manipulating… pic.twitter.com/6omzD423vr
Source: x.com/DrJimFam
Meanwhile, both Meta and Google have announced they are developing their own AI for video creation, but they have not yet been made public.
Impact of AI Video Creation on Digital Marketing
The main impact of AI for video creation in digital marketing is that it will allow companies to generate this type of content without needing a large budget, possibly even at zero cost.
Of course, any corporation could benefit from this, but particularly small and medium-sized businesses that currently lack the resources to produce high-quality videos. Ultimately, this technology will democratize video creation, allowing anyone to produce Hollywood-style creations.
As the arrival of this type of AI is imminent, experts should already start considering how to incorporate it into their strategies.