A large-scale X-ray tomography system is used to acquire both radiographs in real-time, and 3D scans intermittently, during the funnel flow discharge of particles from a rectangular silo for two different levels of packing, i.e. loose and dense, and two levels of side-wall roughness, i.e. smooth and rough. In this way the evolution of the geometry of the flow and stagnant zones have been quantified during discharge for these four idealised cases. No particular effect of the wall roughness on the flowing zone has been revealed for initially dense packing, while an increase rate of the height of the stagnant zone, revealed for the first time from the authors' point of view, was found to get lower as the wall roughness increases. Contrast between flowing and stagnant zones have been enhanced, especially for the loosely packed starting material, by novel difference and variation quantification methods for both 2D and 3D images highlighting differences between flow near the smooth facing walls and the interior as well as concentration variations due to the initial packing conditions. This work shows that X-rays imaging can bring additional insights not revealed by techniques that focus on the flow over the facing walls. (c) 2013 Elsevier Ltd. All rights reserved.