Recently, researchers at the Max Planck Institute in Germany and other institutions in theartificial intelligence (AI)A major breakthrough has been made in the field, where they have successfully utilized theAI technologyRealistic 3D avatars are generated from text, opening up new possibilities for applications such as virtual dress fitting and avatar shape editing.
The DELTA Method: the cornerstone of virtual fitting and shape editing
In the first study, the researchers introduced a method called DELTA, which is capable of creating separate 3D avatars including body and clothing/hair layers. By modeling the body and clothing/hair using different 3D representations, they were able to easily generate these avatars from monocular RGB videos. This decomposition method opens up possibilities for applications such as virtual fitting and shape editing, where users can easily change clothing and hair without affecting the body shape.
The ECA approach: from text to high-quality avatars
Another study proposes the TECA method, which combines a 3D representation of stabilized diffusion and DELTA blending.The TECA method generates high-quality avatars from textual descriptions only, with powerful attribute editing capabilities. The system first generates a facial image as a reference and then gradually adds hair, clothing and other elements. This process significantly improves the quality of the synthesized avatar and enables powerful attribute editing capabilities.
These two studies bring fresh ideas to the field of digital person generation. ByAIalgorithms, the different components of a digital person can be disassembled and recombined to create realistic 3D virtual images. This not only opens up more possibilities for online applications such as virtual dress fittings, but will also have a profound impact on areas such as fashion e-commerce, social platforms and meta-universes. Intelligent editing and generation of digital avatars has opened up a whole new era for future digital experiences.
This article comes from users or anonymous contributions, does not represent the position of Mass Intelligence; all content (including images, videos, etc.) in this article are copyrighted by the original author. Please refer to this site for the relevant issues involvedstatement denying or limiting responsibilityPlease contact the operator of this website for any infringement of rights (Contact Us) We will handle this as stated. Link to this article: https://dzzn.com/en/2023/1252.html
