Ekphrasis is an ancient Greek rhetorical term, which originally referred to a rhetorical device for vividly describing things. By the mid-20th century, it was narrowed down to a literary genre that depicted specific works of art; and after the 1990s, it pointed to the cross-media narrative of language and picture in literary texts. According to Mitchell Foucault, the process of this transformation can be divided into three stages: the indifference, hope and fear of the Ekphrasis. The cross-media narrative of Ekphrasis is fictional in nature, essentially using fiction to overcome the “otherness” of poetry and painting, thereby bridging the gap between poetry and painting, as defined by Lessing, into a language-picture interchange. The cross-media narrative of Ekphrasis involves a triangular relationship between picture, author and reader, and thus forms the mechanism of cross-media narrative: the mechanism of fictionalisation from picture to language and the mechanism of materialisation from language to picture. The former transforms picture narrative into language narrative with the author at the centre, while the latter reduces language narrative into picture narrative with the reader at the centre, both of which form the closed loop of the cross-media narrative mechanism of Ekphrasis.