Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Goal: The goal of this project is to study and propose new schemes of compression of the reflectance function. Description: In a 2D picture, the appearance of objects is strongly influenced by the light that bounces on them. To decouple objects from their illumination and fully capture their intrinsic properties and response to light, we need to use the so-called spatially-varying bidirectional reflectance function (SVBRDF), which is a six-dimensional function that describes the amount of the incoming light that is reflected by a given spatial point of an object, in a specific viewing angle. We currently have several light domes that allow us to capture a high resolution SVBRDF of objects. The complete SVBRDF of a single object has a considerable size (several tens or hundreds of GB) and it can be challenging to manipulate, transmit or even store it without compressing it. Several approaches have been proposed to compress the SVBRDF: the currently most popular ones are model-based and compress the reflectance function separatly for each acquired pixel. As the SVBRDF exhibits a lot of spatial redundancy, it would be interesting to leverage this property and propose more efficient compression schemes (to the best of our knowledge, the only approach that does that is based on radial basis functions). Level: MS Type of work: Research 80%, Coding 20% Skills required: signal processing and information/coding theory LCAV1445050917
Sebastian Urban Stich, Konstantin Mishchenko