Microsoft has developed a modeling approach that can seamlessly produce a 3D virtual avatar from a single 2D image of a human. The diffusion model was developed by a group of experts from Microsoft Research, but it was explained in a paper reported in the journal arXiv.
After this, it applies 3D avatar to develop an augmented reality or virtual reality experience or could be utilized to show a lifelike 3D representation of the person while gaming or for other uses.
A 3D Avatar was Created Using an Algorithmic Method
The machine-learning diffusion concept serves as the foundation for the 3D Avatar Diffusion. Since they are generated, diffusion models may generate new data, like the training data. Although diffusion models had previously been used to build 3D images from 2D photos, the ADM is the first one to be able to create a realistic 3D avatar from just a single 2D image.
The researchers used a dataset with more than 200,000 3D face representations to train the machine. The faces in the collection have a wide range of skin tones, haircuts, and facial characteristics. With just a single 2D image, the ADM might construct a reliable metaverse. As a result, once you understand how the 2D image and 3D face model relate.
Automatically, Generative Models Produce 3D Avatars
This research suggested using a 3D generative model to create 3D digital avatars automatically. Diffusion models and neural radiance fields are also employed to represent them. Because 3D requires so much memory and computing power, making the necessary intricate functionality for high-quality avatars is challenging. Developers propose the roll-out diffusion network (Rodin) solution for this issue.
Bill Gates claims that his old business created the Teams office program, which had 3D avatars and other metaverse-compatible capabilities. Since Microsoft and Mark Zuckerberg are working on Meta, Teams can now be integrated with the latter’s business-oriented workplace social network. Teams might have direct access to the metaverse.
If you want to keep up with the trends of blockchain industry, join our communities on Discord, Reddit and Telegram.