February 16, 2024
Introduction
Sora is a new AI system from OpenAI that can generate realistic video footage from simple text descriptions. Unveiled in February 2024, Sora represents a major advancement in text-to-video generation technology. Where previous models could only produce a few seconds of low-quality, distorted footage, Sora is able to create high-definition video up to one minute in length that accurately depicts the details described in textual prompts.
At its core, Sora is a diffusion model, which means it starts with random noise and gradually refines that noise into a coherent video in several steps This extends and builds on such techniques very similar to OpenAI's previous image generation algorithms such as DALL-E in the video domain. The Sora transformer architecture, which decomposes video into smaller token-like patches in a language model, enables... Read Full Article
