When I talk about the difficulty and the effort related to the letterfitting work I am referring, of course, to the circumstance when the determination of sidebearing and kerning values is carried out manually, value by value, following a procedure where definition and test coincide. Operation that proceeds in an iterative way and in which the final and overall output is the result of thousands of local micro-decisions. A consuming process, therefore, where a constant work of control, of imagination (visualization) and a tiring steadiness of the idea are required of the operator. The frustration, however, beyond the evaluation of the efforts/results ratio and the consideration that a change in planning means starting the whole work again, comes from the simple observation that the process direction, from local to global, is counterintuitive (if not just wrong). The ultimate meaning of the work on iKern has precisely been the elimination of this inversion.
iKern is a mathematical model that describes the white space between the letters. This white space is then reduced to the identification of a possible structure. Where structure is a set of relationships between knowledge, logic and functionality that’s a prerequisite for making use of an abstraction. iKern is also a tool that uses the model to perform letterfitting separating preliminary choices from repetitive tasks, dealing with the former in a smart way and automating the latter. A tool for evaluating the appropriateness of choices by way of the quality of the results. The iKern algorithms identify properties and, in this sense, we can talk of an automatic tool since most of these are directly computable and may be derived from the shapes of characters themselves. This way the model exposes properties that are intrinsic to the glyphs shapes: I call this «to follow the glyphs nature». The properties that can’t be derived become parameters and represent the minimum choice an operator is required to make to guide the process towards the desired result. Acting on parameters implies a different way of thinking: sidebearings and kerning pairs have to be found and not created. The rhythm and the texture have to be designed as a study of balancing visual tensions that the designer has merely triggered. From this point of view iKern is an interface for a system of global decisions where the machine is put in charge of obtaining:
- Coherence: the initial choices are globally and indiscriminately inherited by glyphs and kerning pairs;
- Consistency: the same algorithms and the same parameters produce both autospacing and autokerning;
Not having a theory of reference on which to base the model creation, it has also been necessary to build a new one. It started looking for an answer to a simple but unavoidable question: what is letter-fitting?