Accurate garment fitting in virtual try-on systems remains a significant challenge, particularly when addressing complex body poses, occlusions, and extreme misalignments. To overcome these issues, we introduce a novel approach that enhances both pose adaptability and garment warping by leveraging a global appearance flow estimation model. This work utilises a StyleGAN-based architecture, incorporating a global style vector to capture whole-image context, which improves spatial alignment between garments and body poses. To further enhance fine-grained details, we integrate a flow refinement module that focuses on local garment deformation. Experimental results on the VITON benchmark showcase the superiority of our method, particularly under challenging conditions, establishing a new state-of-the-art in virtual try-on technology. Code is available at: https://github.com/khuong16/Pose-Adapt-VITON-Extends/tree/master .