The evaluation of shading devices is generally carried out using a sequence of shadow-pattern images showing the progression of solar penetration for particular times of the day or year. This approach can reveal when solar penetration may occur, say at the summer solstice, but it cannot give a quantitative measure of the degree and likelihood of solar penetration over a representative period of a full year. This paper describes a new image-based technique to quantify the effectiveness of shading devices. It is founded on predictions of direct solar irradiation using hourly meteorological data for a full year. In addition to numerical output, the technique produces synoptic images that reveal the spatial and temporal variation of solar irradiation. There are no practical limits on the scene geometry and buildings with thousands of individual shading elements can be evaluated. The technique is designed to be both fast and highly scalable making it suitable for the evaluation of a large number of design variants. This is demonstrated in the paper using a parametrically defined model of a complex roof shading system based on the Changi Airport Terminal 3 design. The 3600 fins that comprise the roof shading system were generated using a parametric scheme where the fin orientation has a random component. A total of 42 design variants of the roof shading system were evaluated using the new technique. (C) 2004 Elsevier B.V. All rights reserved.