OpenAI Introduces New Artificial Intelligence Model
OpenAI introduced its new artificial intelligence model Sora. The new AI tool, which creates video from text, surprised everyone with realistic video examples. Here are the details!
OpenAI has done what was expected and introduced a new artificial intelligence tool to create video from text. The artificial intelligence model named Sora, which has already been tested by reliability experts, draws attention with its very realistic results.
Producing 60-second videos, Sora shows how good its new tool is with sample videos shared on OpenAI’s X account. It is still being tested by experts on how it produces results on issues such as hate content, prejudice and disinformation.
The company aims to complete all these vetting processes before making its new tool available to everyone, and to provide a model that delivers reliable results.
New Artificial Intelligence Tool to Create Video from Text
OpenAI shared on its X account some of the images it obtained with Sora, which creates realistic 60-second videos with a text command. Here is the prompt and the video created for the first image it shared:
“Beautiful, snowy and bustling Tokyo city. The camera moves along the bustling city street, following a few people enjoying the beautiful snowy weather and shopping at nearby stalls. Gorgeous sakura petals blowing in the wind along with snowflakes.”
As can be seen from the sample videos, the image quality in the videos is high and retains good detail. It is possible to see that they include complex movements and characters with vivid emotions.
Sora, which has already been described as “terribly” successful, makes the following explanations for its new model: “Sora can create complex scenes with multiple characters, specific gestures, and accurate details of the subject and background. The model understands not only what the user wants on command, but also how these things exist in the physical world.”
Sora, which is being tested for security purposes, will also be opened to artists, designers and filmmakers in the first phase. Feedback from people in this field will be collected through their tests and the model will be further developed.
The company also shared that it has weaknesses as it is still in the testing phase, stating that it may have difficulty simulating complex scenes more accurately and may not be able to understand cause-and-effect examples.
For now, the weaknesses of the model are evident in the videos, such as when a person bites a cookie, but there are no bite marks left on the cookie afterwards.
Sora can convert not only text but also a ready-made image into a video. In these videos, small details and nuances stand out.
