Text-to-Image and Image-to-Image Generation Using Generative AI Diffusion Models

Generative AI has revolutionized image synthesis through advanced deep learning models. Among these, diffusion models have demonstrated superior performance in generating high-quality and semantically aligned images from textual de-scriptions. However, existing systems face limitations in handling long and complex text prompts, maintaining structural consis-tency, and achieving efficient image transformations. This paper presents a diffusion-based framework for both text-to-image and image-to-image generation. The proposed system integrates segment-level text encoding, preference reweighting, and latent diffusion mechanisms to enhance semantic alignment and image quality. By leveraging models such as Stable Diffusion, CLIP embeddings, and Variational Autoencoders (VAE), the system efficiently generates realistic and context-aware images. The results demonstrate improved alignment between textual input and generated visuals, making the system suitable for real-world applications such as design, education, and creative industries.