• Source:JND
Add as a preferred

OpenAI has unveiled its latest AI innovation, the Sora 2 video generation model, almost a year after the release of its first Sora model. This new AI technology comes with significant improvements, allowing users to create highly realistic videos with advanced motion and physics.

Sora for iOS: A TikTok-Style AI Video App

The Sora 2 model is available through a new iOS-only app, which is currently invite-only and available to users in Canada and the US. The app resembles popular short-video platforms like TikTok, Instagram Reels, and YouTube Shorts. Unlike other platforms, every video on the app is AI-generated by Sora 2.

ALSO READ: India’s Online Gaming Bill Comes Into Effect: Esports Recognised, Money Games Banned

Users can upload videos of themselves, and the AI model can recreate their character in new settings with high character consistency. The app also includes a “Cameo” feature, allowing users to generate videos of others using text prompts, essentially creating AI-driven personalised videos.

Major Improvements in Sora 2

OpenAI describes Sora 2 as the “GPT‑3.5 moment for video”, highlighting its advanced capabilities:

- Native audio generation comparable to Google’s Veo 3.

- Realistic physics and complex movements, demonstrated by a gymnast performing flips on a balance beam.

- Fewer hallucinations, meaning objects and reality are less likely to be deformed to execute a prompt.

- Improved controllability, allowing instructions across multiple shots while maintaining consistent world states.

- Best results in realistic, cinematic, and anime styles, making it versatile for different video genres.

ALSO READ: Amazon Unveils New Smart Home Devices, Fire TVs, And Alexa+ With Generative AI

How the App Works

Once invited, users can easily create videos using either the Sora app or website and share them with followers who can like, comment and share them. Furthermore, the Cameo feature adds another level of engagement by enabling users to make videos featuring characters from other videos.

OpenAI emphasises that users must verify their identity before generating personal videos, ensuring that random or celebrity videos cannot be misused. However, the Cameo feature does have some flexibility, which allows for creative applications within the app.

The Growing Trend of AI Video Apps

OpenAI isn’t alone in exploring AI-driven short-video apps. Last week, Meta launched the Vibe feed on its Meta AI app, allowing users to watch short videos created with in-house AI models. This highlights a growing trend of vertical scrolling AI-generated video platforms in the social media space.

Conclusion

OpenAI has advanced AI video generation by creating Sora 2 and its iOS app, making it more accessible, interactive and realistic than ever. Users are empowered with new ways of self-expression through AI-generated videos - potentially signalling a shift in how we create and share short-form content online.