Research
Geometry and Graphics Group
GGG
Inverse Lighting
This problem can be put in the mathematical context of inverse problems, which refer to all the problems where, contrary to what happens with traditional direct problems, several aspects of the scene are unknown. Then, the algorithm is allowed to work backwards to establish the missing parameters.

Reflector Design
The objective is to develope techniques for the design of reflector shapes from prescribed optical properties (far field radiance distribution) and geometrical constraints, which is of high importance in the field of Lighting Engineering, more specifically for Luminaire Design, was proposed. Constraints on the shape imposed by industry needs must be taken into account, bounding the set of possible shape definitions.
Relighting for augmented and mixed reality
Relighting consists of virtually changing the illumination of an existing scene. Often the scene was captured through a camera or viewed on a video, but can also be viewed through augmented reality devices. Challenges become greater when considering complex scenes with dynamic objects and changing lighting conditions, typically an outdoor urban environment. Several linked research topics are associated to relighting:
Capture algorithm
In order to relight one has to capture the scene configuration. This goes from understanding depth and geometry to capturing radiance. At the GGG, we are interested in finding practical methods to capture radiance through the use of high-dynamic range images. Devices and methodologies are created in order to capture complex scene set up with dynamic elements.
Left: Traditional HDR reconstruction. Objects in movement during the capture process are showing ghosting effect in the radiance reconstruction. Right: HDR reconstruction with our method. The object in movement are automatically identified and compensated for in the radiance reconstruction.
Additional credits: Katrien Jacobs (University College London), Greg Ward (Anyhere software).

Publications: Katrien Jacobs, Greg Ward, Celine Loscos, Automatic HDRI generation for Dynamic Scenes, 2005, Technical Sketch, Siggraph 2005.
Katrien Jacobs, Celine Loscos, Greg Ward, Automatic High Dynamic Range Image Generation for Dynamic Scenes, IEEE Computer Graphics and Applications.
Inverse illumination
Algorithms are researched on to retrieve reflectance properties of various materials. Retrieving reflectance properties of object is essential to obtain a consistent and realistic relighting.
Left: One of the input images. Middle: Diffuse reflectance estimation. Right: Relighting using the reflectance estimation and the illumination capture (same illumination conditions as for the picture shown on the left).
Additional credits: Katrien Jacobs (University College London), Anders Hjorth Nielsen and Jeppe Vesterbaek (Aalborg University).
Illumination simulation for mixed reality
The success on relighting heavily depends on the ability of illumination algorithms to adapt to mixed reality. New strategies and algorithms are created to match the requirements of mixed reality which are: interactivity, consistency with the existing scene, complexity of the scene elements in term of geometry and reflectance properties.
Insertion of virtual objects in augmented reality, with shadow consistency. Additional credits: University College London.
Relighting of real scenes. Left: One of the input images. Middle: Relighting with new virtual illumination.
Right: Virtual object insertion with global illumination consistency.
Additional credits: Katrien Jacobs (University College London), Anders Hjorth Nielsen and Jeppe Vesterbaek (Aalborg University).
Publications: Katrien Jacobs, Cameron Angus, Celine Loscos, Jean-Daniel Nahmias, Alex Reche, Anthony Steed, Automatic shadow detection and shadow generation for Augmented Reality, Proceedings of Graphics Interface 2005, May 2005, Vancouver, Canada.