Toward accurate and realistic garment texture transfer with attention to details
Neural Computing and Applications(2024)
Abstract
The categories and styles of garment are constantly diversifying. For designers, it is a pressing issue to evaluate how different fabrics will look on them in a timely manner for users. In this paper, we present a novel garment texture transfer framework from a single person image. Based on the garment model constructed in the person image, we render the texture to the model surface using parallax mapping. To determine the relative positions of the garments when exporting the images, we calculate the contour center moments of the garment mask and the center of mass coordinates of the 3D model and use their consistency to perform position calibration. Finally, we align the rendered garment image with the figure image to obtain the final transfer effect. Experiments demonstrated that our method is robust to different character pose with different garments and background. Qualitative experimental results show that our method accurately and realistically relocates the texture of the garment in the image of the person while preserving the original folds of the garment. Quantitative comparisons with other methods show that our method is optimal in several metrics.
MoreTranslated text
Key words
Texture mapping,Garment model,Position calibration,Image registration
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined