Purpose: Multimodal composing is often romanticized as a flexible approach suitable for all learners. There is a lack of research that critically examines students' perspectives and the constraints of multimodal composing across academic contexts. This study aims to address this need by exploring high school learners' perspectives and experiences enacting multimodal learning in an L2 classroom. More specifically, this study presents key tensions between students' experiences of multimodal composing and teacher/researchers' use of multimodal composition in an L2 classroom setting. Design/methodology/approach: The paper focuses on two multimodal composing projects developed within a design-based implementation research approach and implemented in a high school French class. Multiple data sources were used: observations; interviews; written reflections; and multimodal compositions. Data were analyzed using the critical incident technique (CIT). A critical incident is one that is unplanned and that stimulates reflection on teaching and learning. Methodologically, CIT was enacted through iterative coding to identify critical incidents and collaborative analysis. Findings: Using illustrative examples from multiple data sources, this study discusses four tensions between students' experiences of multimodal composing and teacher/researchers' use of multimodal composition in a classroom setting: the primary audience of student projects, the media leveraged in student projects, expectations of learning in school and the role of a public viewing of student work. Originality/value: This paper problematizes basic assumptions and benefits of multimodal composing and offers ideas on how to re-center multimodal composing on student voices.