INDIVIDUAL DIFFERENCES IN COLOR MATCHING AND ADAPTATION: THEORY AND PRACTICE

Transcription

1 MARK D. FAIRCHILD PROGRAM OF COLOR SCIENCE / MUNSELL COLOR SCIENCE LABORATORY ROCHESTER INSTITUTE OF TECHNOLOGY INDIVIDUAL DIFFERENCES IN COLOR MATCHING AND ADAPTATION: THEORY AND PRACTICE

2 ADAPTATION DEMO Yellow & Blue

3

4

5

6

7

8

9 CMFS & ADAPTATION

10 DO WE NEED INDIVIDUAL MODELS?

11 THEORY Identical Observers

12 THEORY Many CMFs, Identical CATs

13 THEORY Identical CMFs, Many CATs

14 THEORY Many CMFs, Many CATs

15 THEORY Focus of current simulation except each CAT has the same (von Kries) form Many CMFs, Many CATs

16 SIMULATION: CIE 2006 CMFS y 10 20y Relative Sensitivity Wavelength (nm) 8 Sets: 1, 10 20, 40, 60, 80 yo

17 SIMULATION: 3 COLORS Reflectance Factor Wavelength (nm) Gray, Purple, Orange

18 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i

19 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i Tristimulus values of samples under D65 for each observer, i.

20 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i

21 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i TSVs of Ill. D65 for each observer, i.

22 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i

23 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i TSVs of Ill. A for each observer, i.

24 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i

25 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i Ill. A corresponding colors for each observer, i. (These are examined.)

26 COMPUTATIONAL MODEL L CC,A,i = (L sample,d65,i / L source,d65,i ) L Source,A,i M CC,A,i = (M sample,d65,i / M source,d65,i ) M Source,A,i S CC,A,i = (S sample,d65,i / S source,d65,i ) S Source,A,i

27 THEORY M S L M L D65 to A Corresponding Colors 8 Observers Single von Kries Model LMS responses under A S S scale range 50% of L & M

28 THEORY M S L M L D65 to A Corresponding Colors 8 Observers Single von Kries Model LMS responses under A S S scale range 50% of L & M

29 DIFFERENCES ARE LARGE ENOUGH TO LOOK FOR INDIVIDUAL ADAPTATION MODELS

30 DIFFERENCES ARE LARGE ENOUGH TO LOOK FOR INDIVIDUAL ADAPTATION MODELS Just like individual CMFs are interesting

31 SUMMARY STATISTICS: GRAY GRAY L M S Mean Std. Dev Min Max th Percentile th Percentile Percent Std. Dev Percent Deviation Min Precent Deviation Max Goal SEM

32 SUMMARY STATISTICS: GRAY GRAY L M S Mean Std. Dev Min Max th Percentile th Percentile Percent Std. Dev Percent Deviation Min Precent Deviation Max Goal SEM

33 SUMMARY STATISTICS: PURPLE PURPLE L M S Mean Std. Dev Min Max th Percentile th Percentile Percent Std. Dev Percent Deviation Min Precent Deviation Max Goal SEM

34 SUMMARY STATISTICS: ORANGE ORANGE L M S Mean Std. Dev Min Max th Percentile th Percentile Percent Std. Dev Percent Deviation Min Precent Deviation Max Goal SEM

35 PERCENT STANDARD DEVIATIONS LMS A (Corresponding Colors): 5-15% D65 (TSV Specification): 0-10% Observer metamerism alone is about half the variation.

36 PERCENT STANDARD DEVIATIONS LMS A (Corresponding Colors): 5-15% D65 (TSV Specification): 0-10% Observer metamerism alone is about half the variation.

37 THEORY Gray Purple Orange E* E* Overall Goal 0.4 If we could measure corresponding colors with a SEM of about 0.4 CIELAB units, we could determine if there are meaningful differences in CATs among individuals.

38 PRACTICE: CAI & FAIRCHILD POSTER < SEM radius circle 30 Replicates! SEM = SD/N 1/2 Intra-Obs. SEM < 0.5 ΔE* Possible to do better with more replicates.

39 CONCLUSIONS Even with a single von Kries model Individual differences can matter Other models maybe more so Measurements are possible

40 A tree that is unbending is easily broken LAO TZU

41 Thank you.