> Terence Love <[log in to unmask]> kirjoitti 20.2.2016 kello 11.52:
>
...
>
> More specifically, it focuses on how assumptions we make about the continuity of colour spectrums shape the ease, practicality and usefulness of making design theories predicting outcomes in the real world about design activity.
>
> When we assume colour is a continuous spectrum, then in making design theory that includes this we are restricted to using design theory structures and theories that can encompass colour use, design and perception as continuous.
>
> In contrast, when we assume colour is used, designed and perceived as discrete units of individual colour incorporating a range of electromagnetic spectrum , or discrete segments of the electromagnetic spectrum (i.e. each a range of colour that we can regard as the same colour), we can use other design theory structures and theories that do not need to address colour being continuous. Note: there is no assumption that it will be the same bands of the electromagnetic spectrum in each case, only that there will be bands representing each colour rather than colour being continuous spectrum.
>
I don’t believe that discrete/continuous by itself makes such a big difference for ease and practicality of theory. I would rank 3 hypothetical theories by their ease:
1. Theory with 5 discrete colours (easy)
2. Theory where colours are points in continuous space (medium)
3. Theory with 3000 discrete colours (difficult)
The first would lose in practicality and usefulness, it would be just too crude for many tasks. Why I rank continuous theory to be easier than 3. is that if you have a large discrete set, you either have to assume that all colours in there have independent properties or practically treat them as size-0 points in continuous space to calculate their similarity. And you would need similarity to calculate if something that holds for color A would hold for color B at all. If we would assume colors to be independent from each other, color #2999 is whole different case from #3000 and you would need to independently find ’designerly properties’ for all 3000 of them, which would be such error-prone endeavour that the resulting property list wouldn’t be reliable or practical.
What everyone would do in Theory 3 is to assume some similarity metrics across 3000 colours. But when this is done, the colors are treated as continuous and there is no basis for saying that this is easier or more practical than continuous theory. Even when you have discrete dots, e.g. pixels or integers and you start calculating distances (=similarity) between them, you end up with continuous values (floats). Discrete can be translated to continuous when computation requires it, and continuous can be always rounded to discrete.
Theory 1., discrete with small number of colours, would be easiest for computation: we could say that there is no similarity, only identity. Blue is blue and it is not green. Theory could deal each colour as a separate case. Theory with 3000 separate cases would not be practical, easy or useful for humans (nor computers, as colours would be big source of combinatorial explosion, as they would all need to be tested out separately). Theory with 3000 related cases would require continuousness assumption or some more complex method to find if something that holds for color A also holds for color B. (e.g neighborhoods).
For cognition/biology of human colour understanding, I would bank on something like theory 3 with similarities/relatedness enabled (In my mind I can imagine two discretely different shades of green, but not infinitely close and still different), but it won’t be the easiest theory to handle. To make it easier or more practical, I would turn it to continuous theory.
Jukka Purma
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
|