Sora. How will realistic videos from OpenAI change business? | AI in business #94

A few days ago, Google once again caught the attention of AI enthusiasts by announcing Gemini 1.5 just a week after releasing the powerful Gemini 1.0 Ultra model and renaming the chatbot Bard to Gemini.

However, OpenAI didn’t let Google enjoy a strong position in the field of AI for even a day. The creators of ChatGPT revealed an even more exciting solution: Sora, a video generator capable of creating nearly minute-long videos based on textual prompts. OpenAI’s Sora is poised to be a breakthrough tool with the potential to revolutionize the video content creation industry. Yet what makes Sora exceptional isn’t just its technical capabilities but also its wide range of potential applications—from marketing and education to the entertainment industry and everyday communication. Let’s examine how Sora intends to bring about these changes and what prospects it opens up for businesses.

Sora from OpenAI – a new era in video content creation

We’ve come a long way since the days of the first clumsy attempts to create videos using computers. It is already evident that videos created by generative artificial intelligence, even those just six months old, look pretty awkward compared to Sora’s generations. Although Sora, which can create realistic video scenes based on textual instructions, is still in the experimental stage and not yet available to the public, its capabilities are staggering.

Sora offers a wide range of applications that can change the way we create and consume video content. Here are a few examples:

• generating complex scenes with multiple characters or objects,
• animating already existing images,
• editing created videos,
• combining different graphics or videos into one.

These examples show tremendous progress in the field of AI. However, above all, it opens doors to new possibilities in creating video content. Just look at the presentation of a high-resolution video depicting the beauty of a blooming flower:

Source: OpenAI (https://cdn.openai.com/sora/videos/flower-bloom”ing.mp4 or https://youtu.be/UNmqxZoTgsk)

How will Sora be used?

Is Sora just a toy and another tool for creating video content? No. At least according to OpenAI. As the creators of Sora write:

We teach artificial intelligence to understand and simulate the physical world in motion, and the goal is to train models that help people solve problems requiring interaction with the real world.

To generate videos accurately, the model must comprehend the world at a much deeper level than what’s needed for text creation. This entails understanding physics, spatial relationships between objects on the same plane, and the interplay between foreground and background.

Sora will be able to generate:

• video from text – the first and most important innovation of the model,
• video from images – enabling the animation of existing generations created in DALL·E 3,
• video from image and text – allowing for more precise film creation,
• video from video – adding special effects, extending the film, combining two existing recordings, or changing the video style,
• images from text – presumably eventually replacing DALL·E and offering generations with a resolution of 2048×2048.

In the future, Sora could be used to create promotional videos, social media content, or business presentations. It’s a tool that could completely change the way we create and share video content:

Source: OpenAI (https://cdn.openai.com/sora/videos/aquarium-nyc.mp4 or https://youtu.be/3l8wjxjNubE)

Although this colossal step in the development of generative artificial intelligence is exciting, it also raises concerns regarding the risks associated with deep fakes, especially in relation to the US presidential elections. The threats associated with using Sora include primarily:

• Disinformation. Deepfakes, including photorealistic videos generated by Sora, can spread fake news that can sway public opinion and undermine democracy. By creating fake videos that look real, viewers can be manipulated and deceived.
• Violation of privacy. Making Sora available would allow anyone with a computer and internet access to create fake content, including videos that violate privacy and misuse personal data without permission. Probably for this reason, it will take some time before Sora is available to the general public. OpenAI needs to ensure appropriate safeguards.
• Damaging reputation. By using Sora, hackers can impersonate other individuals, organizations, or brands to defame them. By creating the impression that someone is saying or doing things they never did, such videos can significantly harm reputation and credibility. Combining deepfake technology with AI-based voice cloning can, unfortunately, lead to the creation of many highly convincing fakes. As we approach a point where distinguishing a fake from a genuine recording becomes extremely difficult, vigilance, awareness, and critical thinking become increasingly crucial in combating potential harm caused by deepfakes.

Therefore, although Sora’s capabilities are impressive, we must be cautious about their impact on society, create regulations, and take additional steps to minimize their negative consequences.

How to use Sora?

Although Sora is currently in the testing phase and not available to a wider audience, using it appears to be a simple and intuitive process. Users will probably be able to use it as they now use DALL-E 3 in ChatGPT Plus. That is, type text commands, which Sora will convert into short video clips. This offers new opportunities for content creators, marketers, and educators, letting them make engaging, high-quality videos quickly.

How does Sora compare to other AI video generators?

But how does Sora compare to other video generators? For now, we can only speculate how Sora will perform, but based on the description of the tool available on the OpenAI website, we can make some general observations:

• Runway – Sora will work similarly, but with much higher realism. Runway lets you create video and image content using AI, including text-to-video and image-to-image options. It also offers editing enhancements like background removal and element replacement based on text commands.
• Synthesia – the platform operates differently from Sora, as it offers the creation of realistic videos with digital avatars, supporting over 120 languages and allowing for the creation of custom avatars. We don’t yet know how Sora will handle tasks like combining audio or generating speech with the visual material it creates.
• HeyGen – its uniqueness lies in voice cloning and the ability to create an avatar with custom gestures, so its functionality differs significantly from Sora’s offering.
• Stable Video Diffusion – probably Sora’s most serious direct competitor, able to turn static images into video animations, mainly for social media content creators. Its experimental nature, and the fact that it is based on open-access models, raises high hopes in the context of film production and animation.

Sora differs from other video generation tools as it creates highly realistic videos that closely resemble real recordings. Resolution is particularly important here. Sora can make videos with resolutions up to 1920x1080px.

With its deep understanding of language, the model accurately interprets commands. Here, Open AI used the method proven in DALL-E 3. The model first interprets a simple prompt entered by the user and then generates visual content based on its elaborate and detailed version. This allows it to create complex scenes and generate characters that express authentic emotions:

Source: OpenAI (https://cdn.openai.com/sora/videos/closeup-man-in-glasses.mp4 or https://youtu.be/pxkfUDoQg5I)

How can Sora affect the creative industry?

Sora’s potential to transform the creative industry is enormous. Access to this tool for filmmakers and designers brings a new quality to the creation of video content. Sora serves as the basis for models capable of simulating the real world, which could be a breakthrough in achieving AGI (Artificial General Intelligence). At least that’s what its creators, OpenAI, claim.

Since Sora creates realistic moving images similar to those filmed by human hand, it has the potential to significantly change the field of video creation, from training materials to Hollywood productions. Sora will undoubtedly impact:

• Marketing. Sora has the potential to revolutionize advertising campaigns by quickly generating attractive video content based on marketing messages or product descriptions.
• Entertainment industry. Film studies and content creators can benefit from Sora’s ability to create realistic scenes and characters, streamlining production processes.
• Games. Game developers can use Sora to create movies or trailers within games, enhancing players’ experiences.
• Training. The ease of creating high-quality educational materials, including “how-to” instructional videos, can greatly enhance the attractiveness of training and courses.

Source: DALL·E 3, prompt: Marta M. Kania (https://www.linkedin.com/in/martamatyldakania/)

How can companies benefit from realistic AI video?

Using AI in video production offers companies a range of benefits, such as time and cost savings, consistent quality of results, and increased end-product value. OpenAI is taking steps towards ensuring Sora’s safety, including collaborating with anti-adversarial testing teams and developing a classifier to detect AI-generated videos.

Sora from OpenAI opens up new possibilities for creating and consuming video content. From revolutionizing the creative industry to impacting marketing and education, to influencing everyday communication – the potential is immense. As a tool that can completely change the rules of the game, Sora deserves special attention. We look forward to further information from OpenAI, especially regarding when Sora will become available to the wider public. This marks the beginning of a new era in video content creation. The next step is its integration with sound, voice, and 3D models, which will open doors to the metaverse.

If you like our content, join our busy bees community on Facebook, Twitter, LinkedIn, Instagram, YouTube, Pinterest, TikTok.