Thursday, July 20, 2017

the caricature artist seeks to find in a thing that which is abnormal. and indeed it is that which is abnormal which distinguishes a thing from among other things, but how can an artist know what is the average thing among things in order to know what parts of the observed thing are abnormalities? 

an answer is that the abnormalities stand out on their own and they require no frame of comparative reference. However if you're in the business of classifying faces, other things might stand out, just as an auto mechanic looking under the hood, likely isn't immediately struck by how it all looks like a bunch of gobblygook. 

one approach breaks the face up into a finite set of features and classifies each feature from a finite group of shapes. with this approach you run the risk of an awful lot of shoe horning, squeazing things to conform to a shape that is very different from them. but of course if the face has been broken down into enough little peices there will result some resemblance to the total thing.

if that first approach is like classification the other approach is like listing. it basically lays out all the interesting stuff. it's more dependent on finding "a good face." which means a face that the artist sees as having some rare or intriguing aspect to it. and this aspect will be, i think, something which lies outside of the scope of a lot of classifation frameworks.

No comments: