Computer Aided Perception Validation of Tone Mapping Operators in the Simulation of Disability Glare A Masters Thesis Proposal by Charles Ehrlich UC Berkeley Department of Architecture
Jan 12, 2016
Computer Aided PerceptionValidation of Tone Mapping Operators
in the Simulation of Disability Glare
A Masters Thesis Proposal by
Charles Ehrlich
UC Berkeley Department of Architecture
Glare Analysis
Primary Goal
• Provide design professionals with a tool to investigate the visual performance of buildings and the environment
What About Perception?
• Existing methods fail to convey an intuitive understanding of the problem of glare.
Purpose of Image?
• Image as design tool• To convey veridical
information about a proposed building design solution
• Serve as a communication medium for common understanding for the design team
What Makes a Veridical Image?
• Reproduces the perceptual phenomenon that would otherwise be missing due to the limitations of the display medium.
The Fundamental Problem
• Real world luminances: 10-4 to 105 cd/m2 (starlight to sunlight)
• Viewable dynamic range: 1 to 104
• Typical video displays: 1 to 100 cd/m2
What is in an Image?
• 4 Bytes per Pixel– RED– GREEN– BLUE– EXPONENT
• Can store 77 orders of magnitude with 1% accuracy.
Radiance linear Algorithm
Radiance pcond Algorithm
Hypothesis:A rendered image displayed with the appropriate tone mapping algorithm can predict the presence of veiling glare under typical viewing conditions.
Literature Search
Tumblin. Three Methods of Detail-Preserving Contrast Reduction for Displayed Images. Georgia Institute of Tech. 1999.
McNamara. Measures of Lightness Constancy as an
index of the perceptual fidelity of computer
graphics. EU Conferernce on Visual Perception, 1998.
Matkovic. Tone Mapping Techniques and Color Image Difference in Global Illumination,
Technical University of Wein. 1997
Pattanaik, P., Ferwerda, S. et. al. A Multiscale Model of Adaptation and Spatial Vision for Realistic Image Display. Cornell. 1997
Matkovic. 1997. Dissertation: Tone Mapping Techniques and Color Image Difference in Global Illumination.
Pattanaik, P., Ferwerda, S. et. al. A Multiscale Model of Adaptation and Spatial Vision for Realistic Image Display. Cornell. 1997
Tumblin Dissertation
Mcnamara. 1998. Measures of Lightness Constancy as an index of the perceptual fidelity of computer graphics. European Conferernce on Visual Perception.
The Validation Method
• Human subjects are asked to compare:– Scale models of
“complex” scenes with and without high dynamic range
• with– Computer display of
rendered equivalents
scale model
CRT
Confounding Factors
• Scale model versus computer display– Possible to hide the fact that one is a model and
one is a computer display
• Personal sensitivity to glare
• Color perception anomalies
• Corrective visual aids
Experimental Design
• Part 1: Comparison– Method of adjustment– Random ordering of trials– Rate the degree of similarity between scale
model and displayed image on a 5-point scale
• Part 2: Glare– View scale model of historical glare study– Rate perception of glare on a 5-point scale
Sample Size
• 20 volunteers if no funding
• 200 paid subjects, double-blind if funded
Analysis of Results
• Scale Models– Ratings of Similarity
• Glare– Similarity of Glare
Ratings to historical data
Average Scene Brightness
Rated Brightness
Error in Perception
The Solution At Hand
• 3D Visual WYSIWYG
• Advance warning of potential problems– Too little light– Too much light– Disability Glare– Undesirable lighting quality
Veiling Glare
Future Research
• Low light level conditions– Loss of contrast sensitivity– Loss of color acquity– Loss of visual acquity
Loss of Contrast Sensitivity
The Solution in Context
• Building Owners, Architects, Lighting Engineers
• Margin of Safety = over-lighted = waste
• Inappropriate use of Daylighting
Computer Aided Perception System
• Proposed Design• 3D Digital Representation• Simulation System• Display medium (CRT,
paper, etc)• Design Team• A Situation to be Analyzed
or Resolved