OpenAI has presented its new generation neural model for creating videos — Sora 2, Rozetked reported. This model has been significantly improved compared to the first version released in 2024. Now it accurately calculates the movement of objects, preserves object integrity, and can naturally reflect situations similar to real mistakes.
The model precisely executes complex commands and understands various styles — such as cinematic or anime — well. Another new feature is voice synchronization: the model is now able to create dialogues, background sounds, and natural speech.
With the “cameo” function in Sora 2, users can add themselves or their friends to videos using short video and voice recordings.
At the same time, OpenAI launched the Sora app for iOS. Through the app, it is possible to create videos, remix others' work, view the feed, and use the cameo feature. There is parental control for younger users, and the feed is designed to encourage inspiration and creativity.
For now, the app has been launched in the US and Canada by invitation only. The model can be used for free, but in limited amounts. ChatGPT Pro subscribers will have access to the high-quality Sora 2 Pro version.
Read “Zamin” on Telegram!
