Martin Fuchs (Alumnus)

My research interests concentrate on data-driven visual computing, specifically

• Computational Imaging comprises computational photography and computational displays, which combine unconventional optics with algorithms to generate better pictures. For instance, instead of understanding pixels as the “atoms” of images, we can consider radiance distributions on rays – this enables post-hoc control of photographic parameters.
  Recently, we have presented a method that supports the automatic design of faceted mirror systems. Currently, we investigate  spatio-angularly smooth light fields with the  support of the Juniorprofessorenprogramm.
  In the  Transregional Collaborative Research Center 161, we will focus on the quantitative analysis of computational photography.

• Material measurement and appearance synthesis: after previuosly recording scenes with reflectance fields in image space as purely data  driven representations (Fuchs et al. ACM TOG 2007, CGF 2007) and investigating the  synthesis of appearance with digital manufacturing (Fuchs et al. SIGGRAPH 2008, Hašan et al. SIGGRAPH 2010; see also our STAR report by Hullin et al. 2013 ), we currently investigate the efficient control of parametric, procedural models. In our recent publication (Gieseke et al. 2014), we have shown a fast method to achieve this for procedural two-tone textures using photographs as input.

• Interactive Systems: fast GPUs make real-time processing of video streams possible, improving the  recording of video sequences. The growing computing power of mobile devices enable the  capture of 3D geometry with hand-held devices.
  Just as computational photography extends the abilities of conventional cameras,  computational illumination extends the abilities of light sources. Combining both projector and camera in a closed-loop cycle, we can perform in-scene image processing, creating a light source with virtually arbitrary, programmable response (Wang et al., ICCP 2010) for novel human-world interactions.