Incorporating long-range consistency in CNN-based texture generation

Abstract

Gatys et al. (2015) showed that pair-wise products of features in a convolutional network are a very effective representation of image textures. We propose a simple modification to that representation which makes it possible to incorporate long-range structure into image generation, and to render images that satisfy various symmetry constraints. We show how this can greatly improve rendering of regular textures and of images that contain other kinds of symmetric structure. We also present applications to inpainting and season transfer.

References in corpus (3)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Texture Networks: Feed-forward Synthesis of Textures and Stylized Images
• Exploring the Neural Algorithm of Artistic Style

Cited by in corpus (10)

• Stable and Controllable Neural Texture Synthesis and Style Transfer Using Histogram Losses
• On Demand Solid Texture Synthesis Using Deep 3D Networks
• Learning to Synthesize Fashion Textures
• Image Inpainting for High-Resolution Textures using CNN Texture Synthesis
• Dynamic Texture Synthesis by Incorporating Long-range Spatial and Temporal Correlations
• Optimal Textures: Fast and Robust Texture Synthesis and Style Transfer through Optimal Transport
• GANILLA: Generative Adversarial Networks for Image to Illustration Translation
• Non-Parametric Neural Style Transfer
• CAMS: Color-Aware Multi-Style Transfer
• Dialectical GAN for SAR Image Translation: From Sentinel-1 to TerraSAR-X