OpenAI's Sora Turns One Sentence into a 60-Second Video: What Makes It So Good?

Contents

What is ChatGPT, and where is it now?
What exactly is Sora?
What exactly makes Sora so good?
Why are Sora’s videos so detailed?

Sora’s AI breakthrough has stunned the world once again. Its arrival is surely a milestone: it proves, as an actual product, that text-to-video can now produce footage that passes for real. Across social media worldwide, people exclaimed: “Reality no longer exists.”

What is ChatGPT, and where is it now?

Remember ChatGPT? Short for Chat Generative Pre-trained Transformer[2], it is an AI chatbot developed by OpenAI and launched in November 2022. ChatGPT can answer a wide range of questions naturally. It applies generative AI, meaning AI software that can create content such as articles, code, songs and artwork, and it produces text much like a human would write. ChatGPT organizes all kinds of information and finds answers for its users.

Amazon, Shopify, Lazada, OTTO, Wayfair, Walmart, TikTok, Snapchat… from cross-border retail to cross-border B2B, major platforms are all using AI to improve the user experience in shopping guidance and customer service.

Looking back, people were only just discussing and adopting ChatGPT in 2023, and barely a year later, in early 2024, Sora AI arrived, generating video from text. You may be worried again that AI is moving so fast it will replace you. But remember: a time of crisis is also a time of change, and change brings opportunity.

What exactly is Sora?

Since 2023, the AI wave has swept the globe. We are used to generating images with models such as Midjourney and DALL-E, and everything from logos to product photos can be created from an AI prompt. But AI-generated video used to be painfully bad. One of the most famous examples is Will Smith eating spaghetti: in the generated video, Will Smith “stuffs” the pasta in a bizarre way into a mouth that barely looks human, his expression is awkward and unnatural, and the noodles sometimes even pass through his mouth. Overall, it is deeply uncanny.

Sora can understand and generate video that the human eye can take in, and the same technology could one day do the same for hearing, touch and the other human senses. That is why Sora stunned everyone on its first day. Sora’s appeal lies in how lifelike its results are: it goes beyond the limits of traditional video editing and even challenges our sense of what is real.

Sora’s sudden arrival has overturned the views on AI we held just a year ago. AGI may come sooner than expected.

What exactly makes Sora so good?

According to the Sora website, entering the following text into Sora produces the one-minute video shown below:

A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.

For the full video and more examples, visit the Sora website: https://openai.com/sora

Earlier AI text-to-video tools needed constant tuning to produce anything reasonably realistic. Otherwise, you ended up with awkward scenes like Will Smith eating spaghetti above.

Why are Sora’s videos so detailed?

After watching the videos on the Sora website, you’ll surely wonder why Sora’s text-to-video is so good. It’s because OpenAI brought the Transformer into the video’s timeline and into all the related training, which lets the model truly understand far more data. This is what makes Sora unique: it understands the physical world in motion, acting as a general-purpose “physical world simulator,” or “world model.”

Sora uses AI learning to understand how the real world changes and uses computer vision to simulate those changes, creating new visual content. In other words, Sora learns not just videos, and not just the frames in those videos, but the “laws of physics” of the world inside them.

If AI can start learning the laws of physics from “video,” could it one day learn from a “camera lens”? And if so, what would happen if we gave AI a pair of “eyes” and let it roam the world?

Sora is transforming visual expression by turning text into dynamic visual stories. It needs no lengthy production, complex filming equipment or post-production skills: a single passage of text is enough to create stunning visual work.

The world is changing in unimaginable ways. What seemed far away is suddenly right in front of us.

Many of you are probably thinking hard about this year’s plans and how to make them succeed. Looking ahead, the future is full of uncertainty. Technologically, it means that in the near future, it won’t be just one field that is disrupted, but human society and the world as a whole.

Sora’s arrival may be the trumpet call announcing that AGI is truly on its way.

AIM from ESi Technology is an all-in-one LINE marketing platform. It offers solutions for running an engaging LINE presence and understanding your LINE friends, combined with in-depth data analysis, so you can gain friends more efficiently, avoid being blocked, and turn friends into paying customers.

By providing comprehensive, real-time, personalized data, a CDP helps companies adapt to the digital environment, improve operational efficiency, customer experience and competitiveness, and drive digital transformation.

Want to learn more about the customer data platform (CDP) and how to get started? Visit our website for more information.

Open up new channels for growth

See your members in full and build momentum for revenue growth.

Talk to an expert