Адаптивний вибір моделі відбивної здатності поверхні для задач рендерингу

Abstract

In the article, it is proposed the method of adaptive selection of the reflectance model for the tasks of rendering three-dimensional scenes. The peculiarities of calculating the pixel color intensity during the scene object’s surface shading are analyzed. The contribution of the diffuse and specular component to the final color intensity for various materials is considered. The main groups of surface reflectance models are discussed, each of which can be applied only to a certain type of surface. The simplest models of light reflection from diffuse and specular surfaces, more complex light reflectance models that take into account the microgeometry of the surface, and models designed for special types of surfaces are described. The main principles of the concept of adaptive selection of the light reflectance model and the typical stages of its implementation are analyzed. The need to develop new methods of adaptive selection of reflectance models is justified. The features of the proposed method for adaptive selection of light reflectance models are considered in detail. Within the scope of the method, an approach is proposed for choosing the optimal empirical reflectance model to ensure highly productive rendering of the scene. This approach is based on the analysis of the shininess coefficient of the surface. In addition, the recommendations are provided for the selection of light reflectance models for photorealistic visualization of surfaces of a general type, as well as surfaces of a special type, such as fabrics, hair, multilayer materials. At the same time, during highly realistic rendering, it is suggested to use a simple light reflectance model instead of a microfacet model if the surface roughness is small. This approach makes it possible to form a highly realistic image of the object of the scene without excessive computational costs. On the basis of conducting experiments using the BRDF Explorer software tool and using the NMSE metric, the level of surface roughness was determined, where the use of a simpler model is appropriate. The proposed method of adaptive selection of the reflectance model can be used to optimize the graphic systems of highly realistic and highly productive rendering