.jpg)
Generative artificial intelligence (AI) has the disadvantage of not being able to generate high-resolution video that looks realistic. Today this weakness is successfully solved by Sora, the latest generative AI by OpenAI that can generate high-resolution realistic videos of up to 60 seconds. Some of the video examples shared by OpenAI through their official account are amazing.
According to OpenAI, Sora can generate images from various angles and various styles by simply entering the same prom as entered into DALL.E 3. Although the resulting visuals look realistic, OpenAI says Sora still has some shortcomings.
An example is that it is still not really capable of simulating physics and complex scenes. Sometimes it does not understand the instructions by not knowing the difference between left and right. It also does not understand the context like the video above which seems very absurd when it was only instructed to generate a video of a chair being dug out of the ground.
Recognizing the dangers that may exist with Sora now being accessible to users, OpenAI is working to embed C2PA content validation meta data information on every generated video. At this point it is only supported on generative AI generating DALL.E 3 images only.