The Compatibility and Usability Analysis of 3D Modeling Production Pipeline Using Digital Human NeRF-based Technology

The applicability of AI technology is expanding in various content industries, such as broadcasting, film, and gaming, with particular attention being paid to NeRF (neural radiance fields)-based technology, which reconstructs 2D images into 3D models. NeRF can quickly render high-resolution 3D models from a small number of 2D images, significantly reducing the complexity traditionally required in 3D modeling processes. This technology is especially effective when it is necessary to represent complex structures such as human faces accurately, thus reducing both the production time and cost and greatly enhancing the efficiency of content creation. This study evaluates the compatibility and usability of NeRF-based technology in the 3D modeling pipeline for digital human production. Using NeRF-based 3D Gaussian splatting (3DGS) technology, 3D data were generated and applied to an existing graphics pipeline for a suitability assessment. The results demonstrate that 3DGS technology significantly reduces the production time compared to traditional 3D modeling processes while achieving high accuracy in the reconstructed models. However, there were limitations in the detailed representation of hair, eyebrows, and skin textures. Expert validation confirmed that NeRF technology can enhance the efficiency of digital human production pipelines, and with further technological improvements, this technology is expected to be widely applied in various content creation fields.