Real-time Image Generation Utilizing ARM SBC Architecture

Real-time image generation using artificial intelligence has traditionally been constrained by computational limitations on edge devices. This paper presents a novel implementation of real-time image generation on the Jetson Orin Nano platform using Stable Diffusion models. We demonstrate the feasibility of running computationally intensive generative $A I$ models on resource-constrained ARM-based systems while maintaining acceptable performance metrics. Our system processes live camera input and generates corresponding AIenhanced images in real-time. Unlike prior implementations focused on desktop systems, our work provides innovative solutions for ARM architecture, enabling practical applications in real-time environments while achieving frame rates of 2-6 FPS at $512 \times 512$ resolution. Through comprehensive performance analysis, we identify optimal operating parameters and critical thresholds for maintaining image quality while maximizing processing speed. The results demonstrate successful deployment of state-of-the-art generative models on edge devices, providing valuable insights for future edge AI applications in computer vision and real-time image processing. These results highlight the potential of low-power systems to contribute to decentralized highperformance computing frameworks. This opens new pathways for deploying advanced generative models in latency-sensitive, infrastructure-limited, or privacy-critical environments.