Sora AI by OpenAI: Revolutionizing Video Creation with Text Prompts
OpenAI has finally launched Sora, its advanced AI video generation model. The model was unveiled on Monday after months of anticipation, giving the world a glimpse into the future of AI-powered video creation. While a preview of Sora was shared with select users in February, the company introduced an upgraded version, Sora Turbo, capable of producing videos in 1080p resolution for up to 20 seconds. The model is hosted on a standalone platform currently accessible via a website and is exclusively available to paid ChatGPT subscribers, with specific usage limits.
What Is Sora?
Sora is the latest innovation by OpenAI, designed to create motion videos with only text prompts. It was announced in a blog post and represents a major breakthrough in AI technology. The launch was delayed for nine months due to safety and privacy concerns. Since then, OpenAI has made the model's parameters safer and more responsible for release.
Availability of Sora
Currently, Sora is available to ChatGPT Plus and Pro plan subscribers, while the platform is inaccessible for the creation of new accounts to non-subscribers. For the Plus subscription holder, 50 videos can be produced per month at a 480p resolution, but far less allowances for 720p resolution videos are allowed. However, the subscription of Pro offers an extended version at a rate of $200, equivalent to ₹16,970 per month, that extends accessibility for videos in higher resolution and for longer periods along with increasing the usage limit.
Features and Capabilities of Sora
Sora allows users to generate a variety of videos, which can be created in widescreen, vertical, and square aspect ratios. Users can also upload their own images and videos to remix, extend, or blend content with AI-generated visuals. Furthermore, Sora can create videos from scratch based on text prompts.
The platform includes a storyboard feature, enabling users to customize specific inputs for each frame. Technologically, Sora operates as a diffusion model, which helps maintain consistency across multiple frames in a 20-second video. The AI leverages a transformer architecture and borrows techniques from OpenAI’s DALL-E 3 to achieve seamless content generation.
Also Read: Moto G35 5G: India’s Fastest Budget 5G Smartphone at Just ₹9,999
Privacy and Safety Measures
Sora has taken much care concerning its privacy and safety. Training data for the model has been sourced from public domain, partnerships, and proprietary datasets. OpenAI collaborated with Shutterstock, Pond5, among other sources to create special datasets for the model. Moreover, input was sought from AI trainers, red team experts, and employees to fine-tune the system.
OpenAI has, therefore implemented several risk-mitigating safeguards concerning realistic AI-generated videos. Such videos feature visible watermarks and metadata in keeping with the standards of the Coalition for Content Provenance and Authenticity, C2PA. More, the system has included some protections on media upload which concerns persons.
