Research on Graphic Design and Aesthetic Enhancement

This paper proposes a deep learning–based methodology for automating graphic design workflows, addressing critical challenges in traditional 2D-to-3D conversion, image stitching, and visual aesthetics enhancement. To overcome limitations such as alignment errors in low-texture scenarios, poor stitching quality, and loss of artistic intent, the author introduces an unsupervised deep image stitching framework and a parametric 3D reconstruction approach. The methodology can be outlined as follows: (1) an ablation-based loss function optimizes image alignment under large baseline disparities, enabling seamless panoramic generation; (2) a 3D reconstruction network with attention mechanisms enhances spatial depth consistency and design integrity; (3) a multi-scale Gaussian filtering-based aesthetics system improves detail preservation and noise suppression. Experimental validation on 5,000+ pixel-level design samples demonstrates that the proposed method improves workflow efficiency by 35%, reduces noise by 62% in low-texture scenarios, and increases 3D output complexity by 22% in comparison to conventional approaches.