Add to ORKGInternational audienceIn this work, we introduce a dedicated variant of traditional 2D reconstruction in which the input point set (a stippled image) caninclude different regions corresponding to different point densities. Our method converts this input into a layered, vector representation, which includes a main shape plus internal, closed regions, defined by their boundary. Providing such structured output eases subsequent editing and processing, such as generating shaded vector images from the stippled input. To achieve this, we first apply a layered reconstruction algorithm based on the detectedregions in the input point set, and use the output for generating the outer and inner shape boundaries in the stippled drawin