We present a new image analysis technique for measuring the grain size distribution (texture) of the bed surface during flow in a laboratory experiment. A camera and a floating device are connected to a carriage used to take images of the bed surface over the entire flume length. The image analysis technique, which is based on color segmentation, provides detailed data on spatial and temporal changes of the areal fraction content of each grain size at the bed surface. The technique was applied in a laboratory experiment conducted to examine a degradational reach composed of a well sorted two-fraction mixture of sand and gravel. The initial bed consisted of an upstream reach that was characterized by an imposed stepwise fining pattern (the bimodal reach) and a downstream sand reach. A lack of sediment supply and partial transport conditions led to the formation of a static armor in the bimodal reach, which resulted in a more abrupt spatial transition in the bed surface mean grain size. The associated spatial transition in slope led to a backwater effect over the bimodal reach, a streamwise reduction in sand mobility, and so a static armor that was governed by a downstream fining pattern. Although a morphodynamic equilibrium state under steady flow is generally characterized by normal flow, here the partial transport regime prevented the bed from adjusting toward normal flow conditions and the morphodynamic steady state was governed by a backwater. We applied a numerical morphodynamic sand-gravel model to reproduce the laboratory experiment. The numerical model captured the hydrodynamic and morphodynamic adjustment and the static armor well, yet the armoring occurred too slowly. Although the final configuration of the experiment shows features of a gravel-sand transition (i.e., a sudden transition in slope and mean grain size), we are hesitant to claim similarities between our results and the physical mechanisms governing a gravel-sand transition in the field.