Co-occurrence Based Texture Synthesis
This work addresses the need for intuitive and manipulable representations in image generation for users in graphics and design, though it is incremental as it builds on existing co-occurrence statistics and GAN frameworks.
The paper tackled the problem of creating an explainable and controllable texture synthesis model by using co-occurrence statistics, resulting in a method that generates arbitrarily large images with local, interpretable control over texture appearance and enables smooth texture morphing.
As image generation techniques mature, there is a growing interest in explainable representations that are easy to understand and intuitive to manipulate. In this work, we turn to co-occurrence statistics, which have long been used for texture analysis, to learn a controllable texture synthesis model. We propose a fully convolutional generative adversarial network, conditioned locally on co-occurrence statistics, to generate arbitrarily large images while having local, interpretable control over the texture appearance. To encourage fidelity to the input condition, we introduce a novel differentiable co-occurrence loss that is integrated seamlessly into our framework in an end-to-end fashion. We demonstrate that our solution offers a stable, intuitive and interpretable latent representation for texture synthesis, which can be used to generate a smooth texture morph between different textures. We further show an interactive texture tool that allows a user to adjust local characteristics of the synthesized texture image using the co-occurrence values directly.