Inspirational Adversarial Image Generation

The task of image generation started receiving some attention from artists and designers, providing inspiration for new creations. However, exploiting the results deep generative models such as Generative Adversarial Networks can be long tedious given lack existing tools. In this work, we propose a simple strategy to inspire creators with generations learned dataset their choice, while control over output. We design optimization method find optimal latent parameters corresponding closest any input inspirational image. Specifically, allow an user’s choosing by performing several steps recover model’s space. tested exploration methods classical gradient descents gradient-free optimizers. Many optimizers just need comparisons (better/worse than another image), so they even used without numerical criterion nor image, only human preferences. Thus, iterating on one’s preferences make robust facial composite or fashion algorithms. Our four datasets faces, images, textures show that satisfactory images are effectively retrieved in most cases.