The world’s major technology companies are pursuing a new goal: to dominate the field of generative artificial intelligence. One of them is Google, which has brought together all its AI divisions in a very promising proposal, Google Deepmind. Now, the company has unveiled Lumiere, a model capable of generating videos from text.
In addition to the incredible capabilities that generative AI has demonstrated, a large part of its success lies in accessibility. With just a few words, the model will be able to generate an image or text that suits your needs. The same goes for Lumiere: users only have to write what they want to appear in the video and the model will do the rest.
Lumiere also allows you to upload images to transform them into video, modify scenes in real time, or take a visual style as a reference and then imitate it. In addition, Google’s model is capable of generating video with higher quality and realism than its closest competitors. Lumiere’s architecture generates the entire video at once, resulting in more realistic and coherent pieces.
Currently, Lumiere is not available to the public and it is likely that we will have to wait a long time to be able to use it. However, you do have access to a paper that explains Lumiere in detail.