Recent developments have seen D3GA technology innovating in virtual character modeling through the use of Gaussian splatting techniques. This approach constructs realistic and dynamic virtual characters, utilizing 3DGS technology to deliver high-quality images without the need for complex preprocessing. By training individual specific models based on human pose inputs, it enables the driving of various dynamic parts of virtual characters. D3GA's forward mapping method effectively avoids the issue of error accumulation, successfully modeling the color and geometry of virtual characters. The technological innovation, advantages of 3DGS, construction of driven virtual characters, and the forward mapping method all showcase the potential of D3GA in the field of virtual character modeling.