August 2016
Volume 16, Issue 12
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2016
Estimation of illuminant in the white/gold and blue/black "dress" image
Author Affiliations
  • Tomohisa Matsumoto
    Department of Information Processing, Tokyo Institute of Technology, Japan
  • Takuma Morimoto
    Department of Information Processing, Tokyo Institute of Technology, Japan
  • Keiji Uchikawa
    Department of Information Processing, Tokyo Institute of Technology, Japan
Journal of Vision September 2016, Vol.16, 220. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Tomohisa Matsumoto, Takuma Morimoto, Keiji Uchikawa; Estimation of illuminant in the white/gold and blue/black "dress" image. Journal of Vision 2016;16(12):220. doi:

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

It has been argued that individual differences in color appearance of #TheDress may occur as a consequence of color constancy mechanism working differently in individuals. To clarify this argument, we investigated what illuminants observers estimated in the "dress" image, and whether their estimates depended on their color names, i.e., white/gold or blue/black, used to specify the dress colors. In experiments we presented the original dress image and a horizontal stripe pattern (horizontal 19.3 x vertical 13.5 deg, 0.26 cy/deg square stripes) consisting of two colors as test stimuli on a calibrated LCD monitor. Two colors in the stripe pattern were made of blue and black colors averaged in the original dress image in a condition, and in other conditions one of two colors was chosen from a blue-to-white continuum and the other from a black-to-yellow continuum. Observers named two colors in the stimulus using eleven basic color terms and gold, silver and copper. They also selected a color patch, which represented a light illuminating the stimulus, from an array of color patches presented on the LCD monitor. The array consisted of color patches with various chromaticities and luminances. The results showed that when observers used white/gold to name the original dress image, they selected dark bluish illuminants, and when observers used blue/black, they selected bright yellowish illuminants. This tendency to select different chromaticity of an illuminant was also observed when the horizontal stripe pattern was used as the test stimulus although its luminance was not consistent with the case in which the original dress pattern was used. The present results showed strong correlation between stimulus color names and selected illuminant colors, suggesting that different effect of color constancy on illuminant estimation is a determining factor to yield individual differences in color naming of the "dress".

Meeting abstract presented at VSS 2016


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.