CS781 - Colour for Computer Graphics - Winter 2009
Lecture 11
Visual Response to Light
Other interesting observations
- Basic colours
- extension of colour naming
- Boynton's experiment
- Adaptation
- basic idea
- Stiles pi mechanisms
- McCullough effect
- Simultaneous contrast
- completely new colour sensations, such as brown
- probably maintaining colour constancy
- retinex models
Measuring Colour
Two types of measurements
- Visual measurement
- guaranteed to measure things that people see, but
- no guarantee that two people are looking for the same quality
because language is imprecise
- e.g. heterochromatic brightness
- Introduce luminance
- flicker photometry
- easy to create the measurement
- measures to within 1%
- extremely good person to person reproducibility
- except for uncertainty in blue, which is probably related
to aging
- minimum motion tests
- spatial fusion tests
- linearity
- one dimensional subspace of colour space
- Instrumental measurement.
There are two things that we can measure
- psychophysical response to light
- reproduce and improve on visual measurement
- physical properties of light
Measuring physical properties
Almost always energy in the past,
- but now is increasingly photon counting
- photon counting must be calibrated
- by energy measurement, of course
Ultimate calibration is to heat
- Shine a light onto something
- How much does it heat up?
You need to know
- the mass of the material
- the specific heat of the material
- how much heat is lost
- Use this to calibrate a detector
- most sensitive is a photomultiplier
- most common is a solid state detector
- Need to convert energy calibration to power calibration
You now have a detector and a calibration.
- When the meter on the detector reads A (for amps)
- and the wavelength is \lambda
- Then the power of the light source is W (for watts)
Measuring a spectral power distribution
In principle, it is straightforward
- Split the light into a spectrum using
- a prism
- a diffraction grating
- Spectrum can be spread out in
- space, which requires moving detector
- time, which can use a stationary detector
- Get a stream of measurements
- correct for effects of wavelength non-linearity,
- because you are really measuring \Phi(\lambda) \Delta\lambda
Current technology uses an array of detectors, but
- readout is still sequential
Two aspects of calibration are hard
- wavelength: spectral lines used for wavelength calibration
- detector response
- all detectors must be the same size
- far from true in your digital camera
- detectors much be low noise
- far from true in your digital camera
- dark current
- gain
If an instrument is inexpensive they most likely skimped on
calibration.
What can be measured?
Power of light emitted from a source in all directions
- integrated with luminous intensity function
- called luminous intensity
- unit is candela
Power of light enitted from a source in a particular direction
- called luminous flux
- unit is lumen = candela per steradian
- need to talk about solid geometry
Power of light falling on a surface
- called illuminance
- unit is lumen per square metre
Power of light falling on a surface from a particular direction
- called luminance
- unit is lumen per square metre per steradian
Power of light leaving a surface
- called luminous exitance
- unit is lumen per square metre
Power of light leaving a surface in a particular direction
Return to: