- OpenAI just shipped its text-to-video model, Sora, to broad availability.
- It’s limited to ChatGPT Plus/Pro users with resolution and duration caps.
- Sora is powered by a faster, more powerful video generation model.
As a lifelong tech enthusiast who has seen the evolution of AI from a distant observer, I must say that OpenAI’s latest offering, Sora, has piqued my interest like never before. The prospect of turning text into video is nothing short of mind-boggling, and it’s a testament to the strides we’ve made in artificial intelligence.
On the third day of OpenAI’s 12-day “Shipmas” celebration, ChatGPT’s creators unveiled the widespread accessibility of their text-to-video model called Sora. This tool was initially introduced earlier this year in a preview version. The advanced model is capable of producing one-minute videos while ensuring top quality and accurately following the user’s instructions.
At its initial release, users reported significant performance problems, such as difficulties in accurately portraying the physics of intricate scenes and comprehending specific instances of causality and effect. Nevertheless, OpenAI has communicated that the product now operates using a more potent and efficient model called Sora Turbo, which appears to speed up the video generation process.
It’s important to point out that while the latest models of these ships show promising advancements, they do come with significant challenges, as underscored by OpenAI. Specifically, they tend to produce inaccurate physical scenarios and find it difficult to handle intricate tasks extending over prolonged periods.
In the announcement from OpenAI, they share insights about how they’re releasing Sora:
“The version of Sora being rolled out has several constraints. It frequently produces physics that are unrealistic and finds complex actions over extended periods challenging. While Sora Turbo is significantly faster than the February preview, we’re continuing our efforts to ensure the technology becomes accessible to all budgets.”
Let’s unveil our video creation technology so that everyone can discover its potential uses, while we work together on establishing guidelines and protections to guarantee its ethical application as this area of technology progresses.
Every video produced by Sora includes C2PA metadata, which helps identify the video as originating from Sora for transparency purposes. This information can also be used to confirm the source of the content. Although not flawless, we’ve implemented measures such as visible watermarks and an internal search tool that takes into account the technical aspects of each generation to aid in verifying whether the content came from Sora.
Today, we are taking steps to prevent harmful content from being shared, specifically focusing on the prohibition of child sexual abuse materials and deepfakes with sexual content. Initially, uploads will be restricted upon launch, but our plan is to expand this feature to a larger user base as we work on improving our methods for detecting and handling deepfakes. For more information about our safety measures and monitoring practices, please refer to the ‘system card’, along with details regarding our red team testing efforts.
It is our expectation that the initial release of Sora will empower individuals worldwide to delve into innovative creative avenues, share their narratives, and surpass limitations in video storytelling. We can’t wait to witness the masterpieces the world will craft with Sora.
Only ChatGPT Pro and Plus members can access this tool. Consequently, videos created by Sora have a resolution limit of 1080p and a maximum length of 20 seconds.
Users can access the site directly from OpenAI.
As an observer, I note that ChatGPT Plus subscribers can create up to 50 videos in a month at 480p resolution. For those who prefer higher quality, they have the option to produce fewer videos, but with improved resolutions up to 720p per month. Remarkably, the recently introduced $200 monthly ChatGPT Pro subscription offers ten times the usage of Plus users, along with better resolutions and extended video durations.
Next year, OpenAI intends to roll out customized pricing structures based on user categories. Meanwhile, videos generated by Sora are designed to allow remixing or blending, and they’ll be adorned with a watermark for simple recognition of AI-created content, ensuring safety considerations as well.
Read More
- TRX PREDICTION. TRX cryptocurrency
- BCH PREDICTION. BCH cryptocurrency
- ENJ PREDICTION. ENJ cryptocurrency
- DOGINME PREDICTION. DOGINME cryptocurrency
- ETH CAD PREDICTION. ETH cryptocurrency
- ONT PREDICTION. ONT cryptocurrency
- SHX PREDICTION. SHX cryptocurrency
- Nintendo Switch 2 Will Have Backwards Compatibility, Insider Claims
- ETH PREDICTION. ETH cryptocurrency
- XCN PREDICTION. XCN cryptocurrency
2024-12-09 23:09