You can create videos by writing text through artificial intelligence. You can create videos by writing text through Artificial Intelligence: Google launches AI multimodal Lumiere, images can also become videos.
New Delhi3 days ago
- copy link
Now you can create videos directly by writing text with the help of Artificial Intelligence (AI). For this, Google launched its latest AI multimodal Lumiere. Google’s new AI model is a text-to-video and image-to-video model.
It can create videos by making realistic and different types of movements. That is, with it you can not only create a video directly from the text, but also can create a motion video from the image.
In Lumiere, you just need to input text or photo and the AI neural network will generate the video and give it to you. The tool also allows to animate images and create videos in the format of input images or paintings. This model also allows creating special animations and paintings on images and videos.
How does the Lumiere model work?
Google’s new Lumiere adopts a space-time U-Net architecture, unlike the traditional video model, which produces an entire video per frame. In contrast, existing AI video models synthesize key frames one at a time. This new technique eliminates the need to synthesize distant keyframes after temporal super-resolution. With this, stability is easily achieved in the video.
According to Google researchers, the text-to-video generation framework was introduced using pre-trained text-to-image diffusion. The team generated full frame video clips by deploying the space-time U-Net architecture. It includes spatial and temporal modules. It gives good results in video image, video painting and stylized generation.