![OpenAI Sora]()
OpenAI is set to launch Sora, its highly anticipated video generation tool, on Monday, initially available to select users. YouTuber Marques Brownlee shared the news in a recent video, where he provided his first impressions after gaining early access to Sora.
Sora can be accessed at Sora.com, which features a homepage displaying a scroll of recently generated videos curated by OpenAI. Notably, Sora operates as a standalone platform and is not integrated into ChatGPT, OpenAI's popular AI chatbot.
The Sora homepage allows users to bookmark videos for later viewing in a "Saved" tab, organize them into folders, and click on videos to see the text prompts used for their creation. According to Brownlee, Sora can generate videos from uploaded images and text prompts, as well as edit existing videos created through the platform.
One of the standout features is the Re-mix tool, which lets users describe desired changes to a video. Sora will then attempt to incorporate these changes into a new clip. The Re-mix feature includes a "strength" setting that allows users to specify how dramatically they want the changes applied, with higher values resulting in more artistic interpretations.
Sora can produce videos in up to 1080p resolution, but higher resolutions require longer generation times. For instance, generating 1080p footage takes eight times longer than 480p, which is the fastest option available. Brownlee noted that during his testing, an average 1080p video took a couple of minutes to generate, raising questions about how much longer it might take when demand increases.
In addition to one-off clips, Sora features a Storyboard function that enables users to string together multiple prompts to create cohesive scenes or sequences of videos. This feature aims to improve consistency—an area where many AI video generators struggle.
However, Brownlee highlighted some limitations of Sora. It exhibits issues common to generative tools, such as problems with object permanence—where objects appear and disappear without reason—and anatomical inaccuracies with moving limbs in characters.
Sora includes safeguards that prevent the generation of content featuring individuals under 18, violent themes, or potential copyright infringements. It also prohibits creating videos from images of public figures or recognizable characters and watermarks each generated video with a visual mark that can be easily cropped out.
![Media upload agreement]()
Image credit: OpenAI
Despite its limitations, Brownlee found Sora useful for creating title slides in specific styles, animations, abstracts, and stop-motion footage but cautioned against using it for photorealistic content. He remarked that while the AI-generated videos are impressive, they are still easily identifiable as such due to various inconsistencies.
![OpenAI Sora Video generator]()
Video credit
Overall, Sora represents an exciting development in AI-generated video technology, offering unique features while also facing challenges typical of generative models. As it becomes available to more users, it will be interesting to see how it evolves and improves over time.