This paper reviews the landscape metrics used in economic valuations of farmland and forest landscapes. The diversity of landscape indicators used in the literature (land use, landscape composition, distance from landscape assets, seen landscapes: viewshed, objects, shapes) sometimes leads to divergent results. After reviewing the literature, we present a numerical model of the visible landscape, combining satellite images and digital elevation models. Associated with a hedonic model of price valuation, this geographical model offers the flexibility required to test several landscape metrics. The results show that although these have substantial effects on the estimation of the price of a particular landscape seen from a given house, the hedonic prices obtained have rather similar mean values.