GENERATING REALISTIC AI ART: BALANCING CREATIVITY AND CONTROLABILITY

Using diffusion-based generative models, this research introduced a framework for controlling AI art generation. The suggested solution accomplishes a good compromise between user agency and a variety   of   creative   expressions. Work on real-time interactive editing tools, enhancements to bias mitigation techniques, optimisation of generative models for efficient deployment on consumer devices, and integration of multimodal conditioning (e.g., sketch inputs) are all on the horizon.

[1]      Y. Chen et al., “HighResolution Diffusion Models for Image Synthesis,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024.   [2]  M. Kumar et al., “Advanced Denoising Diffusion Techniques for Realistic Image Generation,” IEEE Transactions on Neural Networks and Learning Systems, 2025.   [3]    L. Zhang et al., “TransformerBased TexttoImage          Generation      with CrossAttention,” IEEE Transactions on Artificial Intelligence, 2024. [4]  R. Patel et al., “Conditional Generative Models for Controllable Artistic Image Synthesis,” IEEE Access, 2023.   [5]  H. Lee et al., “Ethical and Responsible AI in Creative Content Generation,” IEEE Access, 2025.   [6]  A. Sharma et al., “Adversarial Training with Style Consistency for AI Art Generation,” IEEE Transactions on Multimedia, 2024.   [7]       T. Nguyen et al., “Explainable Generative Models for Visual Creativity,” IEEE Access, 2026.   [8]     P. Singh et al., “Guided Diffusion Models for Artistic Image Creation,” IEEE Transactions on Image Processing, 2025.   [9]  J. Brown et al., “Semantic Alignment in Multimodal Generative Systems,” IEEE Access, 2024.   [10]  S. Verma et al., “Creative AI Systems with UserControlled Generation,” IEEE Transactions on Artificial Intelligence, 2026.