Enhancing Three-Dimensional Human Digitization using Multi-Layer Pixel Alignment for High- Resolution Imaging

This research explores the application of neural networks to improve the resolution of 3D human images. While existing methods have focused on real-time applications, they often fall short in reconstructing high-resolution 3D images due to memory limitations. The proposed approach introduces a multi-layer architecture that supports end-to-end training, enabling it to focus on fine-level geometry details rather than coarse-level details commonly used in lower-resolution images. This innovative strategy results in significantly improved 3D image reconstruction quality. By addressing the challenges associated with high-resolution 3D image generation, this research contributes to advancements in various fields that rely on accurate and detailed human representations.