The ICT Face Model combines facial scan data in a comprehensive generic model; enabling statistical analysis, and reconstruction of face, head, eyes, teeth and additional geometries standardized for use in real time game- and rendering engines. Our model includes tools for linear and non linear geometry reconstruction as well as submodules for handling texture and detailed shape inference. For information about purchasing VGL's state-of-the-art high resolution 3DMM for commercial use, please contact Kathleen Haase
|Feature||ICT Face Model Light||ICT Face Model Full|
|Base model topology||Yes||Yes|
|#PCA Identity Shape modes||100||200+|
|#PCA Identity albedo modes||0||200+|
|#Expression blend shapes||53||53|
|FBX face rig||No||Yes|
|Albedo to specularity inference||No||Yes|
|Albedo to displacement inference||No||Yes|
Differences between Metahuman and ICT Face Model:
MetaHuman is an artist tool with a few built-in characters and features to play with. It cannot generate novel subjects, only able to change features. ICT face morphable model is a mathematical model based on ~200 subjects, which allows infinite novel subjects to be extracted.
Our work enables personalized avatars/digital doubles creation from images or videos, which MetaHuman doesn't support.
ICT face model can also be used for tracking (e.g., adopted by Nvidia AR SDK).
ICT face model has been built on real scans with real people and bio-information. Each model in the database is Light Stage movie-industry quality.