MAM2015: Eurographics Workshop on Material Appearance Modeling: Issues and Acquisition

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/13212

Browse

Browsing MAM2015: Eurographics Workshop on Material Appearance Modeling: Issues and Acquisition by Subject "I.4.1 [Image Processing and Computer Vision]"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Geometric Accuracy Analysis of Stationary BTF Gonioreflectometers
    (The Eurographics Association, 2015) Havran, V.; Reinhard Klein and Holly Rushmeier
    The accurate BTF data representation requires specialized measurement gantries, some of them designed as gonireflectometers. These consist of an illumination source and a camera mounted on two robotic arms, one degree of freedom possibly achieved by rotation stage which a measured sample is mounted on. While there are several variations of the gonioreflectometer gantry, the principle of all remains the same, positioning directly the illumination and detector on a hemispherical surface over a sample. We analyze the positioning error of such gonioreflectometers. The input parameters are the required spatial resolution of a BTF sample and the distance between the camera used as a detector and the BTF sample. Our analysis confirms that the requirements for mechatronic actuators for the positioning of the sample and arms are very high and near the limit of state of the art technology.
  • Thumbnail Image
    Item
    Quality Assurance Based on Descriptive and Parsimonious Appearance Models
    (The Eurographics Association, 2015) Nielsen, J. B.; Eiriksson, E. R.; Kristensen, R. L.; Wilm, J.; Frisvad, J. R.; Conradsen, K.; Aanæs, H.; Reinhard Klein and Holly Rushmeier
    In this positional paper, we discuss the potential benefits of using appearance models in additive manufacturing, metal casting, wind turbine blade production, and 3D content acquisition. Current state of the art in acquisition and rendering of appearance cannot easily be used for quality assurance in these areas. The common denominator is the need for descriptive and parsimonious appearance models. By 'parsimonious' we mean with few parameters so that a model is useful both for fast acquisition, robust fitting, and fast rendering of appearance. The word 'descriptive' refers to the fact that a model should represent the main features of the acquired appearance data. The solution we propose is to reduce the degrees of freedom by greater use of multivariate statistics.