論文アブストラクト：In this paper, we present a drawing workstation for blind people using a two-dimensional tactile pin-matrix display for in- and output. Four different input modalities, namely menu-based, gesture-based, freehand-stylus and a Time-of-Flight (ToF) depth segmentation of real-world object silhouettes, are utilized to create graphical shapes. Users can freely manipulate shapes after creation. Twelve blind users evaluated and compared the four image creation modalities. During evaluation, participants had to copy four different images. The results show that all modalities are highly appropriate for non-visual drawing tasks. There is no generally preferred drawing modality, but most participants rated the robust and well-known menu-based interaction as very good. Furthermore, menu was second in performance and the most accurate drawing modality. Our evaluation demonstrated direct manipulation works well for blind users at the position of the reading hand. In general, our drawing tool allows blind users to create appealing images.