Facebook already has its own AI-assisted image creation tool, which is how Make-A-Scene works.

Artificial intelligence is becoming more and more part of our lives, and little by little, its rapid growth and wide range of possibilities are forcing many companies to invest a lot of time, money and labor in their development.

This is what we all already know or at least have heard about DALL-E 2, a vastly improved version of its predecessor. OpenAI defines this second version as a new system for creating realistic renderings based on description.

Well, in all companies, including Google, betting on this sector, we find Facebook (Meta). And this is what the company said this Tuesday that it has also developed An AI imaging engine that he hopes will help create immersive worlds in the Metaverse and create high-end digital art.

As they explain Make-A-Scene works very simply. We will provide a sentence and it will go through a transformative model, a neural network that analyzes the words in it and develops an understanding of the context. After that, the AI ​​synthesizes a new image using the GAN (Generative Adversarial Networks) set.

An important argument in favor of these new tools is that they are becoming more precise and sophisticated, enabling all kinds of hyper-realistic images. “Use the potential of AI for creative expression”Mark Zuckerberg stated on the Meta blog Tuesday, “People should be able to shape and control the content that the system generates” keep going.

It doesn’t end here. Allows users to include thumbnails created by these to generate text images, resulting in a 2048 x 2048 pixel image. This allows you to create a more accurate composition.

“It demonstrates how people can use both text and simple drawings to convey their vision with more specificity, using different elements, shapes, placement, depth, compositions and structures.” Zuckerberg said.

At the moment, it is not known when AI will become publicly available.