Enhancing Design Innovation of Realistic Natural Slab Image Synthesis for Industrial Application with Generative AI

Artificial slab manufacture aims at producing natural-looking slabs since customers ask for state-of-the-art materials that still possess the same visual aspect of natural stone. However, the designing process of new styles requires a huge effort in terms of time and resources therefore the employment of Generative Artificial Intelligence enables to improve the number of feasible features and colours combinations in shorter time and with less strain. Leveraging on Generative Adversarial Networks, we obtain a novel framework that, given as input a mask, assigns pixels in the image to a certain feature class, for example background or vein, returns a synthesized slab image, that has realistic look but it is totally synthetic. In order to instruct the network to generate realistic fake images of slabs, we collected a large number of pictures of natural slabs with different patterns and colours and we tagged them according to the class of the features we identify in each slab image. Demonstrating promising potential for industrial applications, our research has successfully produced highly realistic synthesized images of slabs. Future efforts will focus on refining this approach to ensure its suitability for real-world industrial use.