J. Eur. Opt. Society-Rapid Publ. 21, 36( 2025) 63
26 Wagenmakers EJ, Farrell S, AIC model selection using Akaike weights, Psychon. Bull. Rev. 11( 1), 192 – 196( 2004). https:// doi. org / 10.3758 / bf03206482.
27 Burnham KP, Anderson DR, Model selection and inference. A practical information – theoretic approach( Springer- Verlag, Heidelberg, 1998). https:// doi. org / 10.1007 / 978-1- 4757-2917-7 _ 3.
28 Burningham N, Pizlo Z, Allebach JP, Image quality metrics, Encyclopedia Imaging Sci. Technol. 1, 598 – 616( 2002). https:// doi. org / 10.1002 / 0471443395. img038.
29 Khan MU, Luo MR, Tian D, No-reference image quality metrics for color domain modified images. J, Opt. Soc. Am. A 39, B65 – B77( 2022). https:// doi. org / 10.1364 / josaa. 450595.
30 Simone G, Oleari C, Farup I, An alternative color difference formula for computing image difference, in: Proceedings from Gjøvik Color Imaging Symposium 2009.
31 Sheikh HR, Bovik AC, Image information and visual quality, IEEE Trans. Image Process. 15( 2), 430 – 444( 2006). https:// doi. org / 10.1109 / TIP. 2005.859378.
32 Eldarova EL, Starovoitov VA, Iskakov KA, Comparative analysis of universal methods no reference quality assessment of digital images, J. Theor. Appl. Inf. Technol. 99( 9), 1977 – 1987( 2021).
33 Jost-Boissard S, Avouac P, Fontoynont M, Preferred color rendition of skin under LED sources, LEUKOS: The J. Illuminating Eng. Soc. North America 12( 1 – 2), 1 – 15( 2016). https:// doi. org / 10.1080 / 15502724.2015.1060499.
34 Hasler D, Suesstrunk SE, Measuring colorfulness in natural images, in: Human Vision and Electronic Imaging VIII, Vol. 5007, edited by Rogowitz BE, Pappas TN( SPIE, Santa Clara, CA, 2003), pp. 87 – 95. https:// doi. org / 10.1117 / 12.477378.
35 Braun GJ, A Paradigm for color gamut mapping of pictorial images, Thesis, Rochester Institute of Technology, 1999. Accessed from https:// repository. rit. edu / theses / 2860.
36 Sun PL, Morovic J, What differences do observers see in colour image reproduction experiments? Conference on Colour in Graphics, Imaging and Vision 1, 181 – 186( 2002). https:// doi. org / 10.2352 / cgiv. 2002.1.1. art00040.
37 Rousseeuw PJ, Hubert M, Robust statistics for outlier detection, WIREs Data Min & Knowl 1, 73 – 79( 2011). https:// doi. org / 10.1002 / widm. 2.
38 Rigau J, Feixas M, Sbert M, An information-theoretic framework for image complexity, in: Proceedings of the First Eurographics Conference on Computational Aesthetics in Graphics, Visualization and Imaging, Girona, Spain, 2005, pp. 177 – 184.
39 Houser KW, Wei M, David A, Krames MR, Whiteness perception under LED Illumination, LEUKOS 10, 165 – 180( 2014). https:// doi. org / 10.1080 / 15502724.2014.902750.
40 Akinwande MO, Dikko HG, Samson A, Variance inflation factor: as a condition for the inclusion of suppressor variable( s) in regression analysis, Open J. Stat. 05, 754( 2015). https:// doi. org / 10.4236 / ojs. 2015.57075.
Appendix A: Explanation given to the participants
You are invited to participate in our image quality assessment experiment. In this experiment, people will be asked to rank several batches of reproduction of the same image using different gamut. You will be given the original image and asked to rank different reproductions on how good they look to you compared to the others. This is an uncontrolled experiment that can therefore be done under any viewing and lighting condition. The experiment should take approximately 20 minutes to complete. Your participation in this study is completely voluntary. There are no foreseeable risks associated with this project. However, if you feel uncomfortable answering any questions, you can withdraw from the experiment at any point. Your experimental responses will be strictly confidential and data from this research will be reported only in the aggregate. Your information will remain confidential.
You will be asked to rank the six reproductions displayed based on how well they represent the original image.
You have to drag the image on the right panel: on top the one you deemed is the best reproduction and, on the bottom, the one you deemed the worst( Fig. A1). It is only based on your preference. This ranking is subjective and there is no“ right” or“ wrong” answer.
Thank you very much for your time and support.
Appendix B: Metrics tested on previous transmission data
Here, the metric provided in equations( 3) and( 4) is used to predict the preference of observers for the 8 images provided in Figure S1( see Supplementary Information) with the gamuts being: Portrait is the same Bad is the same Printer is adapted to transmission mode as well using Yule- Nielsen model and has way lower lightness 3 gamuts in transmission Figure B1 shows the preference vs prediction graph. The metric is doing decent at separating experimental gamut but ranks the printer gamut too positively despite the lightness being low. The R 2 value here is 0.83, which is better than the 0.76 value of R 2 providedbythemodeledspecifically designed for this question and presented in Supplementary Information Figure S6.
Appendix C: Metrics dispersion for experimental gamuts
Figure C1 presents the disparity of metrics for the 3 experimental gamuts and for 4 images. The images used for this graph are( c) and( o) for low colorfulness images,( m) and( x) for high colorfulness images. For each metric represented in the horizontal axis, there are 4 sub-columns each representing from left to right image( c),( m),( o) and( x), ineachofthosesub-columnsthere are 3 dots representing the values for each of the 3 experimental gamuts. It appears that the most spread metrics are on M11, M16 and MSE, which are not the main ones. M4 and M9 are more useful to distinguish the synthetic gamuts from the rest.
Appendix D: Observers answer dispersion for two types of images
Figure D1 presents the dispersion of the observer ranking for images( c), which seems to have been ranked in the same order by a majority of observers, and image( x), which raised more uncertainty.
Appendix E: Robustness of the model to variations in observer ranking
To assess the robustness of the model to small variations in observer rankings a supplementary analysis simulating plausible sources of uncertainty has been conducted.