Openai introduced the Sora 2, the first major update to video and audio generation models since its debut over a year ago. This new version offers improved tools for sharper realism, synchronized sound, editing and creative control.
This means that Google Veo 3, known as one of the best AI video tools now, has the right competitor.
Sora's “GPT3.5 Moment”
Openai describes SORA 2 as a “GPT-3.5 Moment.” This is a milestone that brings significant advancements in the first release. This model produces videos that follow the laws of physics more accurately, avoiding common flaws such as object morphing and distorted movement, and plaguing previous systems.
SORA 2 can also maintain visual consistency across multiple scenes, allowing users to generate continuous narratives rather than isolated clips. Supports a variety of styles, including film, realistic, and anime, and expands your creative scope.
Improved sound features
In addition to video, SORA 2 provides synchronized interactions and sound effects, addressing gaps in previous models where audio did not match on-screen actions. The upgrade will make the generated clips appear more natural at the same time.
To reduce the risk of misuse, Openai has built strict rules on its model. You cannot use SORA 2 to create videos featuring public figures. Users who want to use portraits to generate content to others must give explicit permission. This can be pulled out at any time via the built-in “cameo” system.
limit
The deployment of Sora 2 will begin today, but access will initially be limited to the US and Canada. In addition to the launch, Openai is introducing a standalone Sora iOS application available through invited-only programs.
For now, the tool is free with “generous restrictions” and offers users the opportunity to test its features at no cost. However, CHATGPT Pro subscribers will receive access to the Sora 2 Pro, a high-quality version of the model.
