Neurally-Guided Texturing for Garment Line Drawings

Abstract

Adding texture to a line drawing is an important process in the production of comics and illustrations. Garment drawings especially often have large deformations with self-occlusions, so deforming patterns is essential for representing realistic garment designs. However, it is currently done manually and requires a significant amount of effort by experts. A possible approach is to infer 3D surface geometry and then apply texture to all 3D surfaces, but it is difficult to represent deep creases using this approach. In this paper, we introduce a "neurally-guided" optimization system for automatically deforming and directly mapping 2D texture patterns to 2D line drawings, bypassing 3D geometry. First, we build a deep neural network to estimate local transformation matrices of texture patterns, called neural-guidance, from line drawings. Second, we build a 2D triangle mesh for the garment and deforms the mesh to obtain the texture coordinates by integrating the local transformations. Our algorithm is effective and easy to integrate into existing drawing systems. We provide several examples to demonstrate the efficiency of our proposed system over previous methods and illustrate the versatility of our method. CCS CONCEPTS • Computing methodologies → Texturing; Shape modeling; Artificial intelligence.

Interactive texture editing for garment line drawings

“…Figure 5 shows some examples of texture editing with our system. The input textured models are generated by using Hashimoto et al 12 Since most existing editing methods, such as FFD, 1,2 do not support discontinuous region handling, such as wrinkles, their results are continuous across the discontinuous regions, giving a flat overall impression (see Figure 2). In addition, while other deformers based on the mesh graph structure 22 can be suitable for considering self‐closing wrinkles, they often create unnatural texture distortions around the wrinkles.…”

“…Since we aim to efficiently “edit” UV coordinates of line drawings, we must set these coordinates to input garment drawings in advance using, for example, a plane projection method. In this article, we utilize Hashimoto's method 12 which fully‐automatically estimates UV coordinates, and prepare some examples of input data. To handle garment wrinkles, the 2D garment models are cut along the crease lines in the input drawings.…”

“…To reduce the bias caused by the difference between the point handles and curve handles operations, this study was conducted twice, once with point handles only and once with curve handles only. First, we provided the subjects with textured garment drawings generated by Hashimoto et al 12 (see Figure 6a, b). Next, we asked them to freely manipulate textures with each method (including each handle) for approximately 15 min, so the total evaluation process took approximately 60 min per subject.…”

“…Unfortunately, these methods do not generate invisible parts, such as the backside and self‐occlusion, and might lead to unnatural texture distortions. Instead of explicit 3D shape estimation, Hashimoto et al 12 constructed a network that estimates local transformation matrices of texture patterns (i.e., $$$ xy $$$‐scaling, rotation, and skew) from line drawings, bypassing 3D geometries. However, since all of these methods are fully‐automatic, it remains difficult to interactively modify the estimation results.…”