Measurement of illumination

Illumination distribution on the atrium floor.

Presentation of illuminance values measured with a Lux-Meter on the floor of the Atrium under identical lighting conditions as the software evaluation.

Table of measured illuminance at sampled points on the floor of the atrium.
 

Four methods are represented, the first one is rendering using BRDF measured attribute surfaces. The second uses diffuse Kd and specular Ks reflectance coefficients averaged from the corresponding BRDF values for a given material. The third method is similar to the second one. This time only averaged diffuse reflectance is considered, and we dub this approach "radiosity".  In the last method, surface attributes were evaluated by visual observation of the materials. We refer to the fourth method as an "artistic approach". For all methods 84 sample points were considered. The simulation results were obtained using the Density Estimation Particle Tracing (DEPT) global illumination algorithm described in [1] which is a part of the Inspirer (Integra, Inc.) system.
 

Summary of comparative data analyses:

• Simulation results of the floor for surfaces rendered with BRDF  attributes.
• Root Mean Square (RMS) Error: 12.2 lux
• Relative RMS Error: 10.5%
• Maximum Error: 34 lux
• Simulation results for surfaces rendered with KdKs averaged from BRDF values attributes.
• Root Mean Square (RMS) Error: 16.0 lux
• Relative RMS Error: 16.8%
• Maximum Error: 40 lux
• Simulation results using radiosity.
• Root Mean Square (RMS) Error: 16.5 lux
• Relative RMS Error: 18.2%
• Maximum Error: 37 lux
• Simulation results for surfaces rendered with "artistic"  attributes.
• Root Mean Square (RMS) Error: 30.8 lux
• Relative RMS Error: 19.4%
• Maximum Error: 77 lux

[1] V.Volevich,K.Myszkowski, A.Khodulev, and Kopylov E.A. Using the Visible Differences Predictor to improve performance in progressive global illumination computations. ACM Transactions on Graphics,  Vol. 19, No. 2, pp. 122-161, 2000.