Fachbereich Architektur Fachgebiet Klassische Archäologie 0.5 1 1.5 2 2.5 3 3.5 4 4.5 n G3 150 160 170 180 190 200 210 220 230 240 250 260 270 0 10 20 30 40 50 θ s [degree ] 9349 150 160 170 180 190 200 210 220 230 240 250 260 270 0 10 20 30 40 50 θ s [degree ] DSF 75. 169 150 160 170 180 190 200 210 220 230 240 250 260 270 0 10 20 30 40 50 θ s [degree ] 70. 0459 150 160 170 180 190 200 210 220 230 240 250 260 270 0 10 20 30 40 50 θ s [degree ] 675 Scatter and its under-explored relevance for the lighting of Roman interior architecture Lighting is one of the key elements of architectural design. Since the in- troduction of window glass in the architecture of Roman baths in the first century, the characteristics of daylit interior spaces depend on its optical properties. The artisanal production processes of Roman window glass lead to air inclusions, uneven thickness and undulated surfaces, which result in scattering of the transmitted light. The scatter effects the quantitative distribution of light over the interior surfaces and the quality of lighting of marble decoration, wall paintings and sculpture. It impacts the human perception of interior scenes and is an unneglectable factor in the understanding of lighting concepts. The amount of scatter in Roman window glass, differences between specimens in terms of era, origin and production method have so far not been investigated, likewise the impact of scattering properties on the light- ing of Roman architectural space. In this pilot study for a more extended investigation of daylighting in Ro- man architecture these scattering properties are quantitatively ascertained for the first time. It particularly investigates: • How methods used in the assessment of modern glazing can be applied for Roman window glass including adverse factors such as small sample size and state of preservation. • Differences between specimens in terms of era, origin and production method, particularly between cast and blown glass (rough-smooth vs. smooth-smooth surface type). • The relevance of scatter for daylighting in Roman architecture. A new method to explore the daylighting of Roman archi- tecture Four pieces (G1-G4) were selected from the archaeological collection of the LVR-LandesMuseum (Bonn) representing a spectrum of provincial Roman window glass findings. 1 The specimens are introspected for their state of preservation by an expert from the Römisch-Germanisches Zentralmuseum (Mainz) (Fig. 1). In the marked area the Bidirectional Scattering Distribution Function (BSDF) of the specimens is measured with a scanning goniophotometer at the Lucerne University of Applied Sciences and Arts. 2 The measurements are repeated for incident angles between 10° and 70°. The generated plots showing the outgoing light distribution in the scattering plane and the an- gle between points of half intensity (FWHM)(Fig. 2). Based on these measurements data-driven transmission models are pre- pared for daylight simulation utilising tools from the Radiance simulation software. 3 The models are applied to a simplistic architectural scene featur- ing a groin vault and a Diocletian window. Lit by a standard CIE sunny sky model 4 set for noon on a summers day the scene is simulated with the four glass models and for comparison without any glazing. From the resulting data, images are rendered mimicking features of hu- man vision such as contrast reduction and glare from bright light sources. 5 (Figs. 3+4) For quantitative comparison a histogram is generated showing the distribution of decadic logarithm luminance values measured from the centre of the scene in two degree steps in any direction (Fig. 5). The unexpected high amount of scattering in blown glass While G4 shows a nearly total diffuse light distribution, G1-G3 show a range of direct transmittance levels. The blown glass G3 shows the highest direct transmittance. Remarkably the angle between points of half intensity of this sample lies between G1 and G2, meaning the blown glass sample has less diffuse scattering over the hemisphere but widens the direct light not less than the cast samples G1 and G2. It can be concluded that the mat de- posit contributes significantly to the results of G4. The glass models signi icantly alternate the lighting of the test scene while blown and cast glass show similar results The measurements are ample to generate valid material models. In the sim- ulation the reference scene with an empty window shows a bright patch of sunlight on the floor and the wall. This patch (d) is the brightest area in the scene, brighter than the sky (w), therefore it is a relevant light source in the scene. The colour distribution in the simulation shows that the reflection A pilot study to introduce a new method to facilitate a new understanding of Roman interior lighting concepts and their relation to artisanal window glass production Measuring and modelling scattering properties of Roman window glass for daylight simulation FWHM=6.37° w w w w d w FWHM=2.90° FWHM=3.98° FWHM=20.07° 0 0 0 0 0 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 n "empty_lwin.histo" 0.5 1 1.5 2 2.5 3 3.5 4 4.5 n G1 0.5 1 1.5 2 2.5 3 3.5 4 4.5 n G2 0.5 1 1.5 2 2.5 3 3.5 4 4.5 n G4 from the outside ground lights the vault, while the sky illuminates walls and floor. Applying the glass models show similar alternation of the scene: • The patch (d) is gone. • The window is the only dominant light source in the scene, therefore the window wall is less bright. • The window is brighter due to scattering of sunlight. Additional light reflected from the ground is recognisable in a gradient at the bottom of the window. This results in a change of the visual appearance the lighted architec- ture: • The scene is illuminated more evenly. • Changing sun position and weather conditions can be expected to have less influence on the lighting. • Windows are visually more important elements in the scene. • The view to the outside is hindered not only by distortion but by scat- tering as well. Conclusion: The new method shows unexpected results and opens a new research ield The adopted method is applicable to the study Roman window glass. The optical properties of Roman window glass have an important influence on architectural lighting. The results clearly call for further study, especially for real architecture as test case. A better understanding is required about the relevance of these results in relation to window orientation, weath- er conditions, wall materials and to Roman daylighting and architectural concepts. Authors Andreas Noback, Technische Universität Darmstadt Lars O. Grobe, Lucerne University of Applied Sciences and Arts Susanne Greiff, Römisch-Germanisches Zentralmuseum, Mainz Jennifer Komp, LVR-LandesMuseum Bonn Franziska Lang, Technische Universität Darmstadt Contact Andreas Noback, noback@klarch.tu-darmstadt.de Literature 1 Komp, Jennifer. (2009). “Römisches Fensterglas: archäologische und archäometrische Untersuchungen zur Glasherstellung im Rheingebiet”. Diss. Aachen: Frankfurt am Main, Univ., Diss., 2007. 2 Apian-Bennewitz, Peter (2014). “Scanning gonio-photometers for asymmetric acquisition of fine-structured BSDF”. In: Eurographics 2014 Workshop on Material Appearance Modeling: Issues and Acquisition. Hrsg. von Reinhard Klein und Holly Rushmeier. The Eurographics Association, S. 1–4. 3 Ward, Gregory, Murat Kurt und N. Bonneel (2012). A Practical Framework for Sharing and Rendering Real-World Bidirectional Scat tering Distribution Functions. Report. LBNL. 4 CIE (2004). 011/E: 2003 Spatial distribution of daylight-CIE standard general sky. 5 Larson, G. W., H. Rushmeier und C. Piatko (1997). “A visibility matching tone reproduction operator for high dynamic range scenes”. In: IEEE Transactions on Visualization and Computer Graphics 3.4, S. 291–306. Fig. 1. Samples of cast and blown glass Fig. 2. Measurement results in the scattering plane (DSF and Full-Widht-Half-Maximum) Fig. 4. Renderings mimicking human perception of the test scene with the glass models Fig. 5. Histograms of decadic logarithm luminance values form the di ferent glass models (w=window) Fig. 3. Rendering and luminance histogram of the test scene without glass (w=window, d=direct light) G1 (Arch.-# 9349) rough-smooth surface t = 3.5 mm few bubbles coriaceous lower side G2 (Arch.-# 75.169) rough-smooth surface t = 4 mm few bubbles flat lower side with small depres- sions, beginning iridescent corro- sion G3 (Arch.-# 70.0459) smooth-smooth surface t = 2 – 3 mm many bubbles beginning iridescent corrosion G4 (Arch.-# 675) rough-smooth surface t = 5 – 6 mm many bubbles flat lower side with small depres- sions, beginning iridescent corro- sion, mat deposit This poster was presented between 3rd - 4th November, 2017 during the congress “Glass of the Cesars @ 30” organised by the British Museum and the Association for the History of Glass.