Abstract: A large number of high-quality tone-mapping operators is currently available. In addition to inherent practical value, comparing their performance is necessary to further advance the field and can provide better understanding of visual realism. In this paper, we show that it becomes very difficult to meaningfully judge relative performance of modern tone-mapping techniques with existing comparison methods and demonstrate that using real environments is crucial in such experiments. We report results of a new study comparing five recent tone-mapping techniques using this approach.