We propose radically new concepts for creating digital and real shapes with the help of computers, considering characteristics of human perception, cognition, and established workflows in art and design. Traditionally, real objects were created and optimized based directly on their visual impression. With the introduction of CAD/CAM, this immediate feedback has been lost, replaced by an engineering pipeline that capitalizes on mathematical representations and accurate machining. We believe to have identified the fundamental problems in this process, and propose research that leads to tools that support creation of shapes by humans and for humans.The research is concerned with data structures and algorithms that support the optimization of virtual and real shapes so that they possess and clearly convey desired features. This will lead to user interfaces for shape design based on features that humans understand and already use for communication. It will also lead to techniques that optimize the geometry of shapes so that the desired features stand out in likely viewing and illumination conditions. We will further extend the optimization to include the illumination, opening up an entirely new way to create the visual world around us.While the research is primarily concerned with geometry, it relies on results in perception, cognitive science, mathematics, and other disciplines, and by means of cross-pollination might lead to fruitful insights across the boundaries of computer science. The resulting tools will help making digital shapes a commodity, with effects on markets, industry, and society similar to what we have experienced for digital music or images.