Notebookcheck Logo

OpenAI reveals Sora, an AI model that can generate photorealistic videos from textual prompts

OpenAI introduces Sora, offering photorealistic text to video generation (Source: OpenAI)
OpenAI introduces Sora, offering photorealistic text to video generation (Source: OpenAI)
OpenAI's latest venture, Sora, breaks new ground in AI innovation by seamlessly translating textual prompts into lifelike videos. Despite some minor imperfections, Sora's ability to generate compelling scenes, from vintage SUVs rallying up mountains to animated monsters, demonstrates its potential in visual storytelling. The model is currently available to a select few people, who are testing it for safety and risks, along with some visual artists and creators for early feedback.

OpenAI has been a trailblazer in the realm of consumer AI, revolutionizing text and image generation, thanks to DALL-E and ChatGPT. Now, with its latest offering, Sora, the company aims to venture into the domain of text-to-video generation, attempting to push the boundaries of AI creativity even further.

Sora boasts the capability to produce videos up to a minute in length, complete with intricate scenes featuring 'multiple characters,' nuanced motion, and 'detailed backgrounds' - all driven by user prompts. The results are astonishingly lifelike, showcasing the undeniable potential of AI in visual storytelling.

In several demonstrations provided by OpenAI, Sora effortlessly brings prompts to life. From a vintage SUV navigating a mountainous terrain to a whimsical scene of a fluffy monster sitting beside a melting candle, the generated videos exhibit a remarkable fidelity to the provided instructions.

However, Sora is certainly not without its imperfections. Some instances reveal minor inconsistencies or unrealistic movements, highlighting areas for refinement, such as in a scene where a cat requests breakfast from its slumbering owner. While the overall scene is convincingly rendered, keen observers may notice subtle movements that deviate from reality as the owner wakes from sleep. Yet, these flaws pale in comparison to the overall capability of the model to produce stunning visual narratives.

One standout example is the depiction of a fluffy monster beside a melting candle. The result appears to be truly splendid, like something out of a Pixar movie. The intricate details, such as the texture of the creature's fur, which would've otherwise been notoriously difficult to render, showcase Sora's prowess in handling complex visual elements.

Presently, Sora is accessible to a select group of "red teamers" tasked with evaluating its potential risks and impacts. Additionally, OpenAI is collaborating with visual artists, designers, and filmmakers to gather feedback aimed at enhancing the model's overall utility for content creators. As Sora continues to undergo refinement and evaluation, it holds the promise of democratizing video creation - offering a powerful tool for storytellers and creators to help bring their visions to life.

Buy the Samsung Galaxy Book3 laptop on Amazon

A still from an animated video of a furry monster generated by Sora (Source: OpenAI)
A still from an animated video of a furry monster generated by Sora (Source: OpenAI)
A still from a video of a vintage SUV rallying up a mountain generated by Sora (Source: OpenAI)
A still from a video of a vintage SUV rallying up a mountain generated by Sora (Source: OpenAI)
As depicted in this still, Sora is clearly not without flaws. Notice how the lady's arm looks nothing like it should as she wakes (Source: OpenAI)
As depicted in this still, Sora is clearly not without flaws. Notice how the lady's arm looks nothing like it should as she wakes (Source: OpenAI)

Source(s)

static version load dynamic
Loading Comments
Comment on this article
Please share our article, every link counts!
> Expert Reviews and News on Laptops, Smartphones and Tech Innovations > News > News Archive > Newsarchive 2024 02 > OpenAI reveals Sora, an AI model that can generate photorealistic videos from textual prompts
Sambit Saha, 2024-02-15 (Update: 2024-02-15)