Deemos Technologies has recently launched ChatAvatar, a text-to-3D digital character creation tool powered by generative AI amid the hype surrounding this emerging technology. With ChatAvatar, users can describe a face using natural language and follow an intuitive interface to create their own unique 3D character.
The software is primarily intended for gamers, who can utilize its technology to generate custom avatars using photos and multi-modal features. Moreover, professionals can also take advantage of the software by replacing conventional asset libraries and minimizing the need to search through existing options.
The software offers pre-made material options that need only minor adjustments, enhancing efficiency and reducing the cost of 3D character creation. Initially designed for assets involving human interaction, this technology will eventually generate any 3D asset based on user requirements. As a result, constructing virtual worlds for gaming, film, and the metaverse will become more affordable. This technology is primarily intended for the gaming and film industries, which spend billions of dollars annually on asset creation.
Although digital humanization has been a popular topic in recent years, there are several issues that must be addressed before it can be widely adopted. “As for the development stage of these digital characters, there is not much difference between now and two or three years ago, as digital avatars are still being used in the process of film and game production rather than functional digital humans.” Di Wu, the founder and CEO of Deemos said to Pandaily. To achieve mass-scale adoption, Di noted, hardware development is crucial for creating high-precision 3D images. Additionally, ethical concerns and development costs must also be considered. Currently, the gaming and film markets are more likely to utilize this technology as it is difficult to determine when conditions will allow for large-scale application.
Deemos Technologies was incubated at ShanghaiTech University and received a pre-A round of funding in 2022, led by Sequoia China Seed Fund and Miracle Plus. Leveraging seven years of facial data from the Plenoptic Stage, a micron-scale face scanning system, Deemos can generate precision-driven assets within one day and parallel processes up to ten at a time.
Deemos aims to further develop its technology for creating “game ready” assets across all categories, including human faces. Unlike other 3D generation techniques that rely on NeRF expression or triangular mesh representation, Deemos’ game ready assets have organized topology and can be bound with high-precision materials. These assets can also be seamlessly integrated into the post-production workflow of games and films.