Sora has arrived, OpenAI’s video generation model now open to all

Cuerpo de la noticia:
OpenAI has taken a significant step in generative artificial intelligence with the launch of Sora, a model capable of creating realistic videos from text instructions. This new technology represents a major breakthrough in digital reality simulation and audiovisual content generation.

The tool allows users to create videos up to 20 seconds long in 1080p resolution, offering unprecedented flexibility. Users can generate completely new content, extend existing videos, or animate static images with a high level of detail and precision.

Sora uses a transformer architecture similar to GPT models, incorporating advanced noise reduction and content generation techniques. The model employs the recaptioning technique from DALL·E 3, enabling it to follow text instructions more accurately when creating videos.

The platform will initially be available to ChatGPT Plus and Pro users, with monthly generation limits. Plus Plan users can create up to 50 videos in 480p resolution, while Pro Plan users will have more options and higher resolution.

OpenAI has implemented important safety measures to prevent potential misuse. All generated videos will include C2PA metadata to verify their origin, visible watermarks, and restrictions to avoid potentially harmful content.

The company considers this initial version an opportunity for society to explore technology possibilities, developing standards and safeguards that ensure responsible use. The goal is to boost new forms of creativity and visual storytelling through artificial intelligence.