Formalize the distinction between motion style and content

Formalize the definitions of motion style and motion content by establishing a strong theoretical understanding of the characteristics that distinguish style from content in character motion, thereby informing more principled motion-style-transfer methods.

Background

STyMo learns motion style from short paired neutral–stylized examples by decomposing style into static and temporal components. Although this decomposition provides practical control, the paper acknowledges that the conceptual boundary between style and content is not yet theoretically well understood.

A rigorous formalization of this distinction could guide the design of more principled representations, learning objectives, and style-transfer systems beyond the empirical decomposition used by STyMo.

References

Finally, while deep learning can capture motion characteristics effectively, we still lack a strong theoretical understanding of what separates style from content. Progress in formalizing these definitions could inform the design of more principled methods.

— STyMo: Fast and Controllable Few-Shot Motion Style Transfer  (2609.04500 - Ponton et al., 3 Sep 2026) in Section 8, “Limitations and Future Work”