And now generative AI is going 3D

2023-06-09 05:04:07

The tools capable of transforming text or images into 3D objects are multiplying: no doubt, generative AI is arriving in the third dimension.

Is there an area that will escape him? After the generation of text, image, video and even music, artificial intelligence is now attacking 3D. With as a last significant example Neuralangelothe AI ​​model developed by Nvidia capable of transforming 2D videos into three dimensions.

Seeing a company valued at nearly a billion dollars embarking on generative AI for 3D videos is probably no coincidence. Already convinced by the potential of the sector, OpenAI had unveiled at the end of December 2022 a tool which transforms textual requests into 3D objects called Point-E. The quality of the images generated is not yet optimal, but it might quickly improve given that AI has already accustomed us to making progress at lightning speed, regardless of the areas.

The quality of the images generated by Point-E is not yet optimal. © OpenAI

In the wake of Point-E, 3D modeling tools have multiplied. These are divided into two categories. On the one hand, those which, like Neuralangelo, transform a 2D image or video into 3D (Wistlabs, Splash…). On the other hand, like Point-E, those who make a text query a three-dimensional object (MovieBot, LeonardoAI…). Not all of these sites have made 3D their specialty. But they took care to add this option to their feature set.

By observing these new tools more closely, one observation emerges: many sites focus on the generation of 3D avatars. This is particularly the case of ChatAvatarof HeadSculpt or even of Go. It remains to be determined how these avatars will be exploited. Will they stay at the stage of entertainment in gadget mode or, on the contrary, will they invest in the professional environment (to present videos in place of a human or to participate in a videoconference)?

Anyway, artificial intelligence is putting back on the front of the 3D scene. Until now reserved for the cinema (which finally abandoned this technology deemed too expensive), three dimensions will soon be within the reach of (almost) everyone, thanks to generative AI.

1686287389
#generative

Leave a Replay