Sora 2: OpenAI’s new text-to-video model brings realistic physics

OpenAI has released Sora 2, a significant upgrade to its text-to-video generator focused on more realistic physics and natural motion. According to reports, Sora 2 reduces unnatural jumps and artifacts in motion-heavy scenes — for example, sports clips where balls and bodies follow more plausible trajectories and interactions.

Key improvements

• Realistic physics and motion: Better handling of motion, collisions and temporal consistency across frames.
• Higher fidelity & longer clips: Cleaner frames, sharper details and support for longer short-form videos.
• Editing & controls: Improved editing possibilities such as targeted re-renders and object adjustments.
• Safety & provenance: Stronger watermarking and metadata to track AI-generated content; new consent/likeness controls.

Availability

The Sora iOS app is initially invite-only in select regions, with plans to expand access. OpenAI offers both regular and higher-quality Pro tiers (Sora 2 Pro) for users such as ChatGPT Pro subscribers.

Concerns

OpenAI’s approach to copyrighted material and content generation has raised discussion in the entertainment industry; the company says it provides opt-out mechanisms and additional safeguards.

For more details and official info, see OpenAI’s Sora page: https://openai.com/sora/

Reported originally by iphone-ticker.de (link removed).

What would you create with Sora 2? Share examples or ideas in the comments.