During this talk I will discuss two approaches for designing human-centric visualization. One approach is based on adoption of visual techniques from traditional visual crafts such as illustration, visual communication, or cinematography. The argument for adopting these techniques is based on the fact that such visual techniques have been used to convey information to humans over centuries, and its effectiveness has been proven by time. They have been developed for human viewer in mind and those techniques that became most successful, effectively utilize visual metaphors to encode underlying information. Successful adoption of these techniques can be achieved in close cooperation with experts on visual communication from traditional visual media.
The approach delineated above results into effective visualization design, but it does not tell us quantitatively how effective a particular visualization design really is. The second approach for effective human-centric visualization design is therefore based on evaluation of visualization, where the outcome from the evaluation step tells us in numbers how much of intended information has been conveyed to the human correctly and how error-prone is a particular visualization design. This information can be fed back into the visualization design to improve visualization effectiveness. The outcome after several iterations of visualization evaluation and redesign is a visualization technique that conveys the intended information close to optimum and the quantitative characteristics originating from the evaluation serve as a basis for prediction of how much of the intended information will be interpreted by the humans correctly, and what error rate is to be expected during the interpretation step of the visualization pipeline.