3D scanners are professional, structured-light systems that capture geometry with sub-millimeter accuracy, enabling the creation of virtual garment digital twins. However, these high-resolution scans produce dense and heavy meshes that overwhelm performance and storage capacities in immersive environments. This study presents a scalable workflow to generate lightweight yet realistic 3D garment models for Metaverse applications. The novelty lies in the proposal and validation, via 20 padded garment case study, of a process that uses Artec Studio 19 for scan consolidation, Blender for retopology and texture baking, and Unity Engine 3D for deployment, while examining acquisition limitations (e.g., occluded folds, reflective materials) and quantitatively characterizing the trade-off between visual fidelity and performance. Starting from meshes of approximately 13 million triangles, we apply a 10% decimation followed by quad-based remeshing to achieve \(\sim \) 30,000 quads and produce FBX assets of only 18–25 MB without perceptible quality loss. Side-by-side renderings in Blender and Unity confirm indistinguishability, and these asset sizes satisfy the 40 MB guideline for a dedicated Metaverse platform, supporting under 4 s of Web-AR load times. Best-practice guidelines detailing acquisition parameters, post-processing workflows, and optimization settings provide a reproducible framework for creating digital fashion models in Metaverse applications.