The Vision and Graphics Lab (VGL) is a world-renowned research group focused on state-of-the-art, photorealistic 3D morphable face modeling and authoring using deep learning. In 2020, we developed the first real-time system that can produce a complete, clothed human body including textures using only a single 2D view to digitize a high-fidelity 3D body model. The VGL also developed a framework that takes a single scan as input to generate a set of personalized facial blendshapes, dynamic and physically-based textures, as well as secondary facial components like teeth and eyeballs. (See our Publications page.)

In addition to the above mentioned research accomplishments, the Vision and Graphics Lab has had the pleasure of working with top-tier talent, studio executives, commercial research groups, and production teams working in film, games and new media. We deliver high-resolution facial reflectance data from our light stage for film projects such as Free Guy, Aladdin, Jungle Book, Logan, Bad Boys 3, Avatar 1,2 & 3, Thor Ragnarök, and Wonder Woman 1984 (just to name a few).

Our partnerships within the entertainment industry help us uncover new challenges in the virtual digital human space. These insights inform our research goals and push us to investigate new and exciting capabilities. If you are interested in tech transfers including light stage scan services and data, or light stage system transitions, please get in contact.

Processed Data Deliverables
Based on Company’s selection, USC can deliver the following Processed Data for each expression:

  • Scan*

    • Obj ~ 5 million polys or greater
    • Maps (6K)
      • Normals Tangent
      • Displacement
      • Diffuse Unlit
      • Diffuse Normal (r,g,b)
      • Diffuse Albedo
      • Ambient Occlusion
      • Specular Normal
      • Specular Unlit
      • Specular Albedo
      • Single Scatter
    • Camera lineups
    • Undistorted diffuse and spec image from each camera
    • Per camera calibration file (or data to generate maya camera) of lineup of scan to undistorted Image of the subject

    *We also register the scans to a uniform topology(see Emily data), so the geometry and textures are corresponded across all scanned expressions. We can also wrap the scans to client provided topology if discussed before processing the scans.

  • Optional capture list, subject to COMPANY needs and discussion among the parties:

    • 16k face mask that separates each area captured during scan session.
    • 16k micro-displacement map merges all captured face areas into one displacement map synthesized over the neutral displacement.*
    • Subject multiple facial expressions, Polarized- Z-Swirl Pattern.
    • Subject multiple facial expressions, Non-Polarized – Z-Swirl Pattern.

    *For information about how to use the micro displacement map together with other texture maps, see: and

  • Calibration material

    • Macbeth DC Chart
    • Grey Ball

Footer With Address And Phones