User-Controllable Multi-Texture Synthesis with Generative Adversarial\n Networks

2019/04/09 by Aibek Alanov, Max Kochurov, Alanov, Aibek +9
Computer Science · Engineering · #3D Shape Modeling and Analysis #Computer Graphics and Visualization Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.1904.04751

openalex publication_date 2019/04/09 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28

Abstract

We propose a novel multi-texture synthesis model based on generative\nadversarial networks (GANs) with a user-controllable mechanism. The user\ncontrol ability allows to explicitly specify the texture which should be\ngenerated by the model. This property follows from using an encoder part which\nlearns a latent representation for each texture from the dataset. To ensure a\ndataset coverage, we use an adversarial loss function that penalizes for\nincorrect reproductions of a given texture. In experiments, we show that our\nmodel can learn descriptive texture manifolds for large datasets and from raw\ndata such as a collection of high-resolution photos. Moreover, we apply our\nmethod to produce 3D textures and show that it outperforms existing baselines.\n

Related

• Texture Representation via Analysis and Synthesis with Generative Adversarial Networks
• Texture Synthesis with Spatial Generative Adversarial Networks
• Diversified Texture Synthesis with Feed-forward Networks
• Texture Networks: Feed-forward Synthesis of Textures and Stylized Images
• Learning Texture Manifolds with the Periodic Spatial GAN