Lightcuts is a framework for computing complex illumination at a point by exploiting light coherence and approximating contribution from many light sources while bounding the introduced error. It scales logarithmically to the number of light sources and can significantly improve the quality of the illumination without sacrificing speed. Originally proposed as an offline technique, we investigate where the bottlenecks of Lightcuts are and how it can be applied to improve the quality and scalability of an existing interactive global illumination algorithm. Further work would focus in improving convergence of the algorithm by computing view-dependent solutions.