|"Rules: Logic and Applications" 2nd Workshop, Dec, 2019|
|Slides as PDF||Contact|
I'm a cartoonist, interested in how I transform visual information when drawing. Good drawing, I suggest, can be regarded as optimisation, in that it translates the visual language of reality to that of pen-and-ink as closely as possible given the limitations of the second language.
It can also be regarded as a generalised inverse, undoing the ill-effects of these limitations as much as is possible. I give two examples, both well-known to artists: adding information about depth in order to compensate for lack of stereopsis, and deleting texture in order to restore tonal balance.
The idea also applies to other kinds of translation, and I shall briefly relate it to analogical reasoning and to translating poetry.
I believe it is worth mining art for other kinds of image transformation — morphisms, as category theorists would call them — which "know about" aesthetic structure, and I shall illustrate this with examples from cartooning, Cubism, and Cézanne. A catalogue of such transformations might provide a more knowledge-based approach to computer art than given by the current fashion for deep learning. I shall demonstrate this with an example of automatically enhancing cartoons.