The photo-realistic rendering of scenes showing natural phenomena requires skilled graphic designers not only to produce a convincingly good-looking image but also to convey physical plausibility. This is especially important in industrial context, where a modelled scene showcasing a product has to approximate the actual environment of a product as closely as possible, e.g. in automotive industries. In this thesis, new techniques to measure natural phenomena are presented in order to provide new or verify existing models for rendering the physically plausible image. In contrast to other approaches, the measurement is performed using nonconventional methods: an ellipsometer is employed to capture the specular reflectance with respect to the polarisation behaviour, a transmissive screen attached to a glass tank is imaged to capture underwater reflectances, and the Microsoft Kinect, a motion capturing device, is used to detect the gas flows around objects. The results are the verification of existing, physically plausible models for commodity metals, an enhanced reflectance model for materials immersed in transparent media with known refractive index, and the reconstruction of two-phase gas flows around occluding objects.