TEXT-IMAGE GENERATION-STABLE DIFFUSION

TEXT-IMAGE GENERATION-STABLE DIFFUSION

Unnati kolte, Samprada Kale, Priti Yamkar, Sakshi Deshpande

Department of Computer Engineering, Savitribai Phule Pune University,

 Pune – 411007

___________________________________________________________________________________________________________________________

Abstract: The text to image conversion task has long been a challenging problem in the field of computer vision. Recent advancements in generative models have led to the development of various text-to-image synthesis techniques. In this project, we explore the use of the Stable Diffusion framework for text-to-image synthesis. The Stable Diffusion model is a diffusion-based generative model that can produce high-quality images with fine-grained details. We propose a novel approach to generate images from textual descriptions using the Stable Diffusion model, which involves learning a joint embedding space for the text and the image domains. Our approach enables the Stable Diffusion model to generate highly realistic images that are closely aligned with the given textual descriptions. We evaluate our proposed approach on several benchmark datasets, and our experimental results demonstrate the effectiveness of our proposed method for text-to-image synthesis.