Even in the early history of live-action film and cartoon animation, many works blended both media to insert hand-drawn characters into the real world. Films today use photorealistic 3D rendering to achieve seamless blending between real and virtual worlds. However, the dependence on 3D models and physically based rendering has limited the artistic freedom and expressiveness of hand-drawn cartoons. Upcoming films and augmented reality applications introduce cartoon and comic book characters into the real world, but leave behind the stylised shading in the original media. This PhD topic proposes research into blending characters with non-photorealistic rendering (NPR) into real scene, while preserving these stylisations. Related work addresses issues into realistic composition, NPR techniques, and perception in NPR media. However, little is known about how abstracted stylisations depict presence in real scenes. Preliminary research has measured seamlessness in realistic composition. The thesis will extend the methodology to measure perceived blending of multiple cartoon stylisations in a range of real scenes. New live-action compositing methods will be developed for both 3D modelled and hand-drawn cartoons.