This study applies the philosophy of technological mediation to understand how vision-based systems used for civil structure condition assessment transform input of images from cameras into output of structural response. The objective of this study is to understand the mediating role that a vision-based system plays between their users and properties of civil structures that are unperceivable in a great part by humans. A case study of a specific application of such a system is conceived as a responsive digital material with substrates (image frames consisting of millions of pixels) and traces (modelled time-histories to be interpreted by their users). Built-in technological selectivities determine possible differences between modelled time-histories and existing civil structures in reality and attribute to epistemic uncertainties. Modelled time-histories amplify a person’s experience of a certain aspect of the monitored civil structure while simultaneously reducing experiences of other aspects of the structure. Therefore, to prevent human blindness, engineers have to judge outputs of condition assessment systems in the light of specific pre-theoretical shared agreements and routines. Understanding the complex interplay between technological selectivities and different types of human blindness may increase understanding of how these systems can be used successfully in civil structure condition assessment.