Based on M.A.K. Halliday's theories of metafunction in systemic-functional grammar, Gunther Kress and Theo van Leeuwen constructed the theory of visual grammar with three dimensions: representational meaning, interactive meaning and compositional meaning. In the wake of repeated practices, such a theoretical framework, which has been frequently applied in multimodal discourse analyses, has almost heralded an academic consensus, yet also precipitating researchers into a rigid pattern. However, this article gathers six posters from three different animations, that is, “Nezha Conquers the Dragon King” (1979), “The Legend of Nezha” (2003) and “Ne Zha” (2019), to anatomize the diachronic development of visual grammar in multimodal discourse. It concludes that posters in the new era lay more emphasis on narrative representation in representational meaning; make more use of indirect eye contact, long social distance, low angle and high modality in interactive meaning; and take on more personalized processing of information value, salience and framing in compositional meaning. Such a comprehensive analysis of visual grammar in multimodal discourse can not only foster a more profound understanding of the themes and connotations of Nezha's posters, but also signal an aesthetic evolution under changing times, thereby providing inspiration for contemporary poster designs and cultural transmissions.