Abstract:We propose a novel method to achieve robust and wrinkle-realistic clothing reconstruction from a single RGB image. Our approach successfully exploits both advantages of implicit function methods and explicit clothing generative models. The former allows to capture high-frequency wrinkle details and clothing appearance from a single image, and the latter provides reasonable clothing shape prior and structured topology. To utilize the implicit function method to obtain implicit clothing, the key is that we need to automatically segment the clothing region from the reconstructed results and solve its depth ambiguity problem. Therefore, we design a pixel-aligned segmentation method to achieve automatic and complete clothing segmentation. It relies on a lightweight clothing mask network and a simple but effective segmentation strategy. To deal with the depth ambiguity problem, a depth correction function is introduced to significantly remove the uncertainty in the depth direction and recover the correct clothing shape. Further, we utilize explicit clothing generative models to provide reasonable shapes and topologies. It first infers an initial explicit clothing template based on generative models. Then, the implicit clothing is fitted by this explicit template to capture high-frequency deformation, resulting in wrinkle-realistic and structured clothing results. Extensive experiments have proved the validity of our approach with the visual effect and geometry accuracy both reaching the state-of-the-art level in clothing reconstruction from a single image.

Textile Wrinkle Shaped from Image

Dynamic recognition of fabric wrinkle from image

Encoding Localized Strain History Through Wrinkle Based Structural Colors

An Image-Based Texture Mapping Technique for Apparel Products Exhibition and Interior Design

Mapping and color separation algorithm of woven photograph CAD

Unique Aspects of a Shape Memory Polymer As the Substrate for Surface Wrinkling.

A new texture morphing method for visual presentation of textile product

A Novel Approach to the Detection of Facial Wrinkles: Database, Detection Algorithm, and Evaluation Metrics

A Novel Approach to the Detection of Facial Wrinkles

Investigation of 3D surface profile reconstruction technology for automatic evaluation of fabric smoothness appearance

Experimental Research of Membrane Wrinkle Based on Photogrammetry

Improved genetic algorithm application in textile deformation simulation

Recognition Algorithm of Fabric Based on Tamura Texture Features

Photorealistic Facial Wrinkles Removal

A Hybrid Rendering Algorithm for Textile Objects

Computer Simulation of Brocade Based on Modeling Structures

Image based high realistic visual pattern effect achieving method

Automatic Pose and Wrinkle Transfer for Aesthetic Garment Display

DeepWrinkles: Accurate and Realistic Clothing Modeling

Research On 3d Textured Dress Geometry Modelling

Wrinkles Realistic Clothing Reconstruction by Combining Implicit and Explicit Method