In visual experiments the human sensory response to light stimuli seems to generate an ‘additive colour value’ that is linearly additive.
The response may be visualized using the terms of Lightness Chroma and Hue.
The nature of colour identity and predicting the colour of additive colour mixtures seems to be an intrinsically three dimensional (3D) task.
The CIE Standard Observer 3D colorimetric model works well and is widely used for industrial match prediction.
The problem is that there are several well documented cases of predictive failure using such three-dimensional models see:- RS Berns 1997A generic approach to color modeling- - Color Research & Application: Vol. 22 pp 318 et. sq.
The task is to both explain and model such failures using more spectral dimensions if necessary.
David, I am not quite sure what your question is about. What exactly do you want to match or predict? Spectra, tristimulus values or colour appearance? After a quick scan of Berns's article I admittedly have not understood very much of it, but it seems to me he dealt with the question how to match tristimulus values or spectra from different colourants (under a given illumination except for self-luminous colours). There is a short discussion of metamerism. Is that what you mean with "predictive failures"? To model such failures seems not too difficult, but to avoid them is a challenge.
You're perfectly right, The problem is the colour computing based on Newton. In 1987, When I made sculptures I was in front of a Problem. For my sculptures I used as material the painting in a space. When I used black and white , it worked very well. I was able to built a virtual object observed in the space.
But When I used colour it became a mass of unformed colours.
So I reorganised the colours in the space and observed they did not follow the newton spectrum and where dynamic together. I organised the colours in What I called the théorie of the Dynamic colours, never published.
Laurent Nottal I met, used this for his "théorie of the dynamic chaos".
I met Jacques Lewiner Who wanted me to publish. but the death of André Marion with whom I worked let me alone in my little laboratory.
I have developed a new colour synthesis and developed "the colour dynamic mecanisme". I need to work with a new peer to be able to publish. But, for now, I have no time to find one. I was burgled by many firmes and institutes each time i contact them so , in 2000 I decided to stop all contacts and developpe in my laboratory.
If you are interested by my works look at My web site www.albert-jobbeduval.com.
i usse some little tools i built for art work imaging.
If you are interested in a colour mixing model , you can consider a colour appearance model (CAM) maybe. Although Berns' generic model includes more complicated reflectance functions (e.g. corrections for refractive index discontinuities), CAMs have a wider scope in terms of visual phenomena (e.g. saturation, brightness changes, adaptation).
Thank you all for your interesting queries. I have spent many years studying and developing colorimetric models concentrating mainly on colour identity and colour match prediction, for example as specified by CIE Standard Observer X,Y,Z tristimulus values. Virtually all the colour difference and colour appearance models use CIE XYZ values as a scalar basis and I have shown (see my PhD thesis) that as such some key ambiguities and failures of definition are introduced into the modelling process. Further, CIE colorimetry thus specified is a strictly three dimensional model and therefore other than by modelling metameric (or conditional) visual equivalence by tristimulus sum, it cannot be used to go any deeper into metameric modelling. The initial mission would be to characterise and classify all the relevant ambiguites and failures of definition and the wider goal would be create new models of them for publication and consideration for general use. See also the added PDF file.
It's quite long and quite complexe to go from the state of the art in colour computing to the colour reality. So, I will write a little draft to explain my way of research and send it as soon as it's finished.
I tried to write the draft but I'm so busy and it is very difficult to speak about a model I didn't publish without giving some fundamental informations.
For example:
In 1996, during a lunch with C Lahannier who was the head of research in C2RMF,
I draw the model of an hyperspectral camera using interference filters.
Christian asked me at which wavelength i think to scale the filters I gave him the scale but with the central wavelength because I used interference filters.
Christian, asked a firm to build the camera developed the software with polytechnique , ENST.
I was humiliated to not be the head of the research groupe as he told me when he asked me to draw the camera.
I was hangry when I discover they develop the camera without me.
In 2000 I met the guy who built the camera and laughs a lot when I saw that all the filter's scales were some lowpass!
The more terrible is that yesterday I met the PhD with whom I work in physicochemical analysis. He bought a new camera on the same model.
But still with the same lowpass scales.
When I give some sheets of information not fondamental.
They burgle it make enormous mistake and say that what I do don't work!