Four knowledgeable referees support acceptance for the contributions, notably the proposed geodesic distance between elliptically symmetric distributions over CNN filter outputs as a way to interpolate between textures and the evaluation of the method by psychophysical similarity and neural responses. I also recommend acceptance. However, please consider revising your manuscript– being more forthright in discussing the psychophysical results in comparison to other studies and also the non-ellipticalness of natural textures as pointed out by a reviewer.