The popular video generator from OpenAI is introducing its new flagship model, Sora 2. It offers more realistic motion, more accurate shots, convincing sound effects, and a new cameo feature that opens the door to creativity, but also to ethical questions. What can Sora 2 do? And why is it attracting criticism as well as enthusiasm?
Sora 2: Accurate simulation of the real world and synchronized dialogue
Sora 2 is the latest video generator from OpenAI. Using only a text prompt, it can create hyperrealistic clips that are almost indistinguishable from reality. It has a better understanding of light, space, scene depth, and camera movement itself.
According to the demonstrations released so far, Sora 2 is far more accurate, handles intricate details, understands body mechanics, supports synchronized dialogue and convincing sound effects, while its scenes maintain logical continuity in movements and expressions. In short, the new generation pushes the limits of imagination and makes it very easy to create realistic scenes and shots that are sometimes indistinguishable from reality.
First vs. second generation of Sora
The first Sora video generator model was launched in February last year and became an instant sensation. However, users often criticized it for “losing logic” in longer sequences and distorting reality and objects. Objects and scenes frequently changed shape or disappeared, disrupting the video's flow and credibility.
Sora 2 was launched on September 30 this year and surprised users by being able to maintain consistency across storylines and generate seamless audio, unlike its predecessor. Sora 2 was specifically trained to simulate the real world, which means it can understand even highly complex activities such as Olympic gymnastics. Nevertheless, OpenAI states that the Sora 2 generator still makes mistakes, though less frequently than its predecessor.
Where to find Sora 2
You can try Sora 2 as usual at sora.com and now also in the Sora app for iOS. For now, however, it is in testing mode for selected creators, and the app is available by invitation only. Once the app becomes widely available, however, it promises an entirely new video experience.
Sora App: Edit videos or become part of them
According to OpenAI, the new Sora app is intended to become an interactive social network where users will create and share their videos and have the option to modify or edit existing sequences.
Another new feature will be cameo, which allows you to insert yourself into any AI video and become part of an AI-generated reality. In the app, you simply create your own avatar, requiring only a scan of your face and voice. Sam Altman himself demonstrated this feature in the Sora 2 promotional video.
Deepfakes, safety, and copyright
Like any powerful and realistic technology, Sora 2 is also raising concerns. Some experts believe that the model's hyperrealistic capabilities and cameo feature could lead to misuse and the spread of deepfake videos and disinformation. However, OpenAI has introduced safety measures to prevent such incidents.
Creating an avatar using the cameo feature requires more than simply uploading a photograph; you must record your moving face together with your voice. In addition, videos will include a visible watermark and a C2PA metadata signature.
OpenAI also wants to prevent the spread of harmful content and protect children from excessive use of its platform. It has therefore not only developed a safety guide, but also made parental controls available.
The launch of Sora 2 has also caused an uproar among television and film studios. Social networks are being flooded with videos featuring copyrighted characters and footage. However, OpenAI has stated that creators and rights holders must actively opt out if they do not want their works to be used. The first company to deliberately exercise this option was Disney.
There is no doubt that the arrival of Sora 2 marks the beginning of an entirely new era of video creation, while also introducing major challenges in safety and rights protection. If OpenAI can maintain a balance between creativity and safety, Sora 2 could mark the beginning of a revolution in AI video.
Source: OpenAI, CNBC, CNN Business, ČTK
