OpenAI can now turn your words into ultra-realistic videos. That said, the availability of the feature will be open to experts and creative. Whether it will be open to others in the future is still yet to be seen.
OpenAI Ultra-Realistic Videos
OpenAI, the AI startup renowned for creations like ChatGPT and Dall-E, has introduced a groundbreaking text-to-video model named Sora, marking a significant advancement in generative AI technology.
Comparable to Google’s Lumiere, Sora elevates the capabilities of text-to-video generation, albeit with some distinctions. Notably, Sora surpasses Lumiere in its capacity to generate videos up to 1 minute in duration.
The emergence of text-to-video technology has sparked a competitive frenzy among leading players in generative AI, including OpenAI, Google, and Microsoft. With the sector projected to amass $1.3 trillion in revenue by 2032, companies are vying not only for market dominance but also to captivate consumers who have been captivated by generative AI since the advent of ChatGPT just over a year ago.
OpenAI Announces Sora
OpenAI’s announcement regarding Sora outlines its intended audience, which includes “red teamers” versed in areas such as misinformation, hateful content, and bias. These experts will rigorously test the model to assess its performance under adversarial conditions. Additionally, OpenAI seeks feedback from visual artists, designers, and filmmakers, aiming to incorporate insights from creative professionals. Such collaborative efforts are pivotal in addressing concerns surrounding the potential for convincing deepfakes, a prominent issue in AI-generated imagery and video content.
Furthermore, OpenAI’s decision to share Sora’s progress publicly reflects its commitment to transparency and its desire to provide a glimpse into the evolving landscape of AI capabilities. By engaging with external stakeholders and soliciting feedback, OpenAI aims to foster an environment of collaboration and innovation in the field of generative AI.
Strengths and Benefits
Sora, OpenAI’s latest innovation in generative AI, boasts an impressive ability to interpret long prompts, including examples spanning up to 135 words. In a sample video shared by OpenAI, Sora showcases its versatility by generating a wide array of characters and scenes, ranging from people, animals, and whimsical monsters to intricate cityscapes, serene landscapes, zen gardens, and even depictions of New York City submerged underwater.
This remarkable capability is attributed, in part, to OpenAI’s previous advancements with models like Dall-E and GPT. Notably, OpenAI released Dall-E 3, a text-to-image generator, in September, with CNET’s Stephen Shankland lauding it as a significant improvement over its predecessor, Dall-E 2. Moreover, OpenAI unveiled its latest AI model, GPT-4 Turbo, in November, further enriching its suite of cutting-edge technologies.
Sora leverages Dall-E 3’s recaptioning technique, which generates highly descriptive captions for the visual training data. This technique contributes to Sora’s ability to produce detailed and nuanced visual interpretations based on textual input, enhancing its capacity to generate diverse and compelling video content.
“Sora is able to generate complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background,” the post stated. “The model understands not only what the user has asked for in the prompt, but also how those things exist in the physical world.”
Weakness and Drawbacks
OpenAI acknowledged that Sora has shortcomings, such as difficulties in accurately portraying the physics of intricate scenes and comprehending cause-and-effect relationships.
“For example, a person might take a bite out of a cookie, but afterward, the cookie may not have a bite mark,” the post revealed.
And for those who still rely on making an “L” with their hands to discern left from right, Sora offers little relief, as it also occasionally confuses left and right.
Sora’s Widespread Availability
While OpenAI didn’t disclose a specific timeline for Sora’s widespread availability, it emphasized the importance of taking “several important safety steps” beforehand. This includes ensuring that Sora meets OpenAI’s existing safety standards, which prohibit extreme violence, sexual content, hateful imagery, celebrity likeness, and the unauthorized use of intellectual property.
“Despite extensive research and testing, we cannot predict all of the beneficial ways people will use our technology, nor all the ways people will abuse it,” the post revealed. “That’s why we believe that learning from real-world use is a critical component of creating and releasing increasingly safe AI systems over time.”
MORE RELATED POSTS
- Arizona State University Partners with OpenAI to Use ChatGPT in Classrooms
- OpenAI Superalignment Fast Grants 2024 (Up to $2,000,000 Grant) – APPLY NOW
- Experts and Critics Not Happy About the New App Store Tax
- Users Can Now Use the DALL-E 3 AI Image Generator inside Bing Chat
- OpenAI is Granting Access to DALL-E 3
