We developed a non-invasive method to measure light interception and radiation-use efficiency (RUE) in thousands of maize (Zea mays) plants at the PHENOARCH phenotyping platform. Different models were interfaced to estimate (i) the amount of light reaching each plant from hemispherical images, (ii) light intercepted by each plant via a functional-structural plant model, (iii) RUE, as the ratio of plant biomass to intercepted light. The inputs of these models were leaf area, biomass and architecture estimated from plant images and environmental data collected with a precise spatial and temporal resolution. We have tested this method by comparing two experiments performed in autumn and winter/spring. Biomass and leaf area differed between experiments showing a high G×E interaction. Difference in biomass between experiments was entirely accounted for by the difference in intercepted light. Hence, the mean RUE was common to both experiments and genotypes ranked similarly. The methods presented here allowed dissecting the differences between experiments into (i) genotypic traits that did not differ between experiments but had a high genetic variability, namely plant architecture and RUE (ii) environmental differences, essentially incident light, that affected both biomass and leaf area, (iii) plant traits that differed between experiments due to environmental variables, in particular leaf growth.