Exploring interactions between shading and view using visual difference prediction

For a given view content, how well a view can be seen through different facade layers, or 'view visibility', depends on both the properties of the facade and the incident lighting. This dependency makes view visibility challenging to measure and quantify, which makes it difficult to understand how the benefits of a view through a window may be preserved by partially occluding facade layers. With the objective of relating the optical properties of different dynamic shading systems to what will be visible through these systems, this paper leverages existing vision models to synthetically analyse a range of facade and view scenarios. The primary outcome is that one such model, the HDR-VDP-3, usefully responds to the scenarios tested. The response is as expected for parameters with known impacts on visibility, such as direct and diffuse transmission, and free viewing area. It is also sensitive to view features such as high-contrast edges, low-contrast details and composition. Moving forward, these findings should be used to confirm the fundamental attributes of human visual perception across shading systems and, relying on human subjects, calibrate them to models' responses. This can help guide future research into preserving the 'key' qualities of a view.