Vinker et al. (2022) | CLIPasso: Semantically-Aware Object Sketching

The work by Vinker et al. (2022) differs from the other key papers referenced in this book. The CLIPasso model is able to convert a (raster) image of an object into a drawing in vector image format (SVG). The required abstraction skills are obtained from the CLIP model which is able to distill the semantic concepts from both images and sketches.

This paper was presented at SIGGRAPH 2022.

Available resources at a glance

Vinker et al. 2022 paper

Fig. 53 Screenshot of the CLIPasso paper by Vinker et al. (2022)