CSAIL Publications and Digital Archive header
bullet Research Abstracts Home bullet CSAIL Digital Archive bullet Research Activities bullet CSAIL Home bullet

link to publications.csail.mit.edu link to www.csail.mit.edu horizontal line


Research Abstracts - 2007
horizontal line

horizontal line

vertical line
vertical line

Context Sensitive Color Rendering from LED White Lighting

Maria R. Thompson & Una-May O'Reilly

Previous research has proven that integrated lighting controls can be highly effective at reducing the amount of electrical energy used for fluorescent lighting in commercial buildings. It is also known that advanced control strategies that use occupancy sensors to reduce lighting of unoccupied spaces can reduce energy consumption by at least 35% compared to already efficient systems [1]. Such advanced control systems can potentially enable further improvements in energy savings performance if coupled with the sophisticated controllability features afforded by Solid State Lighting (SSL) technology, such as light emitting diodes (LED).

The components of a white multi-chip LED are essentially Red, Green and Blue (RGB) chips. The color properties of such a white light, are a function of the proportion of RGB used for the mix, and they have crucial influence on the appearance of objects and spaces lit by that light. The main color properties of white light are called Color Temperature and Color Rendering. Color Temperature, CT is the appearance of light emitted from a white light source. It defines how warm (yellowish), or how cold (bluish) a certain white light may appear [2, 3, 5]. Color Rendering, CR defines the color appearance of objects when illuminated by the white light source. Different light sources affect the color appearance of objects differently (figure 1). Light bulbs, such as incandescent and fluorescent bulbs, come from factory with a fixed value for CT and CR. Multi-chip white LEDs allow for modulation of those properties.


Figure 1: Contrast between high color rendering (right) and low color rendering (left).The brightness and color temperature of the two images are equal

LED efficiency can be increased by strategically adjusting the values of CR [4]. While adjustable light output (dimming) is well understood as an effective method to save energy in unoccupied spaces, LED lamp efficacy can be increased up to 30% by strategically lowering the color rendering. The notable point is that while modulating CR, the light level and CT can stay constant and therefore the appearance of the light source stays the same. How the objects are rendered under the light source is a key, ensuing question. If they can be rendered without noticeable change in quality, we could manipulate the lighting quality in a room without people noticing it. At no time would the space appear dark, uninviting or potentially unsafe (because light level (i.e. brightness) and CT are not changed). Additionally, if it were possible to smoothly transition between different CR modulations, it would not create any apparent distraction. Everyone wants good lighting. And, no one wants irritating lights that poorly respond to activity in a manner that is supposed to be "intelligent". Our hypothesis is that the most valuable intelligent lighting will be completely invisible.  

Our first step toward this potential exploitation of context sensitive color rendering with white LEDs consisted of an investigation of the perceptual impact of dynamic modulation of CR with color-mixing white LEDs. In the Fall of 2005, we conducted a series of visual experiments which allowed us to understand if and how reduced CR would be acceptable for users. Observers evaluated a sequence of LED light settings and compare these to a reference source, an incandescent bulb. While we modulated Color Rendering, light levels and color temperature were held constant and the same as the reference source. We were able to demonstrate: (a) Perceived difference of different CRs; (b) Visual tolerance to reduced CR; (c) Correlation between reduced CR and various color reflectance (high and low saturation) to define ranges of colors that suffer the most from reduced CR.  

A year later, on a second research phase we followed up our booth experiments with a full scale mock-up pilot study. We asked the same questions concerning object rendering quality but we manipulated the light in the context of a much more realistic everyday environment, and the research focused on the evaluation of color perception under highly automated lighting control strategies. The search was primarily inspired by the same idea: bring the lighting of unoccupied spaces to minimum color rendering to reduce energy consumption without altering color temperature or light levels. If successful, the control technique could minimize peak hours lighting energy waste, and potentially enable up to 25% of power reduction.

On this second phase a series of visual experiments were performed based on subjective assessment of color changes under modulated color rendering from white LEDs. Subjects evaluated a sequence of white spectra generated by a multi-chip LED system in comparison with a reference source (incandescent). Again, the color temperature and illuminance of all light sources were held constant, while each of the LED spectra had a different color rendering Index (CRI). The main research question was if people would notice when the lighting in the LED room was automatically transitioning through different color rendering. The hypothesis was that a range of color distortions would be negligible for the visual perception of several architectural settings. Visual tests were carried out for central and peripheral vision, using two full scale mockup rooms, where subjects compare the appearance of a dinning scene under LEDs and under incandescent lighting. Results confirmed the fundamental hypothesis, showing that the majority of subjects did not detect the color changes in their periphery while the same color changes were noticeable with direct observation. The outcome include fundamental guidelines for how to extrapolate the experimental results into real life and apply the data to architectural settings. Hypothetical architectural scenarios were presented and the potential for energy savings was discussed. The results also provided us with guideline data (and a deeper understanding) that indicate human perceptual tolerance to color rendering manipulation.  

The data is essential in the design of a software control system with LED ceiling panel hardware, an LED actuation network and a sensor network to support novel energy saving lighting control strategies. Such a control system could achieve savings by intelligently linking the current activity in a space to the output from modular LED ceiling panels. It would incorporate distributed sensing algorithms to allow context sensitive modulation of light properties, particularly color rendering.  

Figure 2: Subject facing illuminated booth openings and color charts.

Modulation of light quality properties (such as CRI) is still an open question in the field of solid state lighting. The first phase of our project was presented at the 6th Light Research Office Symposium in Light and Color, Lake Buena Vista, FL, February 2006 [6]; and the second phase was presented at the CIE Expert Symposium in Visual Appearance, Paris, Octoberm, 2006. A more detailed analysis of the combined studies were the core of Maria Thompson’s PhD thesis [8]. We received support from Osram Sylvania who provided us with the LEDs.


[1] F. Rubinstein. Long-Term Results from the Lighting Controls Demonstration at the Watergate Building in Emeryville, CA. Final Report for Pacific Gas and Electric Co., May, 1991. LBL Report LBL-28793.

[2] J. Schanda. The Concept of Color Rendering Revisited. Color in Graphics Imaging and Vision. April 2002.

[3] M. S. Shur, A. Zukauskas. Solid State Lighting: Toward Superior Illumination. Proc. IEEE, vol.93, NO. 10, pp.1691-1703, Oct. 2005.

[4] Y. Ohno. Color Rendering and Luminous Efficacy of White LED Spectra. Proc. SPIE, vol.5530, Aug 2004.

[5] F. Schubert, J. K. Kim. Solid State Light Sources Getting Smart. Science 308, 1274-1278 (2005).

[6] M. Thompson, U.M. O'Reilly. An Investigation into the Perception of Color under LED White Composite Spectra with Modulated Color Rendering. Proceedings for the 6th Light Research Office Symposium in Light and Color, Lake Buena Vista, FL, February 2006.

[7] Thompson M., O’Reilly UM, “An Investigation into the Perception of Colors under Dynamic Modulation of Color Rendering in Real Life Settings”, Proceeding of the CIE Expert Symposium in Visual Appearance, Paris, October 2006.

[8] Thompson M., “Psychophysical Evaluations of Modulated Color Rendering for Energy Performance of LED-based Architectural Lighting”, PhD Thesis, MIT, February 2007.



vertical line
vertical line
horizontal line

MIT logo Computer Science and Artificial Intelligence Laboratory (CSAIL)
The Stata Center, Building 32 - 32 Vassar Street - Cambridge, MA 02139 - USA
tel:+1-617-253-0073 - publications@csail.mit.edu