<![CDATA[This study documents and analyses a practice-based collaboration in which Indonesian and Malaysian designers used generative artificial intelligence (AI) to synthesise shared batik motifs, and examines how faithfully the source motifs survived that process. Adopting a practice-based / research through-design approach, the study brought together practitioner researchers from three Indonesian and two Malaysian institutions. Six iconic source motifs Parang, Truntum, and Mega Mendung (Indonesia) and Bunga Kerak Nasi, Pucuk Rebung, and the Tiger Orchid motif (Malaysia) were combined through a structured prompt protocol using two text to image platforms (Midjourney and Bing Image Creator) and refined in Adobe Illustrator. The resulting nine motif outputs and two garment visualisations were assessed through iconographic visual analysis and collaborative expert appraisal across four dimensions: cultural fidelity, cultural distortion, design innovation, and design-stage resource efficiency. The analysis shows that generative AI was effective for ideation and cross tradition synthesis but unreliable for fidelity. Only the discrete, radially repeatable Truntum was reproduced with high fidelity, whereas the directional, field-organising geometric motifs Parang and Pucuk Rebung were the least faithful; because this weakness appeared in motifs from both traditions, it reflects a structural limitation of the technology rather than a national bias. Colour was the least controllable parameter, and culturally faithful results emerged only through a human corrective stage, positioning AI as a partner in human AI co-production. The study contributes a replicable cross-national workflow, an analytical mapping of motif fidelity, and a reframing of the two traditions from contested ownership toward shared authorship. ]]>
