Modelling Geometry and Appearance

Creating digital models of physical objects is an important concern in the field of computer graphics. A digital model of an object combines a description of the surface geometry with a description of the appearance of the object. These two descriptions can be extremely simple as, say, in the case of a sphere which reflects incident light evenly in all directions, and they can be of great complexity as is often the case when we create models of real world objects.

Our research into digital geometry and appearance is method oriented and application motivated. We develop methods for modelling geometry and appearance, creating digital twins of actual objects or processes, and analysis of geometry and materials. These topics are discussed below in greater detail.


  • Biomedical shapes, skeletonization and shape descriptors.
  • Separation of appearance characteristics into those that pertain to microgeometry and those that pertain to physical properties.
  • VR for visualization of large scale geometry.
  • Analysis by synthesis (inverse rendering) for fine tuning models.


  • Interplay between geometry and appearance at the microscopic scale.
  • Generative (aka procedural) methods for creating synthetic 3D models.
  • Shape representations with specific properties such as polar annular meshes (a mesh skeleton co-representation).
  • Mechanical and architectural structures: how we create designs that are manufacturable at large scale as well as meshes for load bearing structures.
  • Interactive modelling, including work on sculpting and other types of shape creation using desktop tools, VR systems, and tablet devices.
  • Methods for tracking topologically adaptive deformable interfaces.
  • Emergent phenomena in complex, dynamic systems composed of simpler entities.
  • Digital master: modelling the shape and appearance of a product before manufacturing with applications to quality assessment.

Digital twins


  • Acquisition of geometry and optical properties that describe object appearance.
  • Additive manufacturing and control of shape and appearance.
  • Simulation: how will something look? How will fluids interact with some geometry?


As a part of our dissemination, we have developed or contributed to a number of apps and open source codes.

  • The GEL library contains numerous tools for geometry processing and analysis. The library is developed in C++ but contains Python bindings.
  • The DSC (Deformable Simplicial Complex) library is a framework for topologically adaptive deformable interface tracking.
  • During the course of several student projects, we have developed an interactive box modeling system: The VR Modeling System
  • Several apps have been developed in projects together with the TopOpt group at DTU Mechanical Engineering (TopOpt Shape, TopOpt, TopOpt3D)
  • Example implementations of methods for computing the scattering of light (LMabs, dirpole).
  • Interactive illustrations of basic mathematical tools for procedural modelling (onb, noise).




Jakob Andreas Bærentzen
Associate professor
DTU Compute
+45 45 25 34 14


Jeppe Revall Frisvad
Associate Professor
DTU Compute
+45 45 25 33 58