Pixel-Aware Stable Diffusion (PASD) aims to achieve realistic image super-resolution and personalized style transfer. By introducing a pixel-aware cross-attention module, PASD enables the diffusion model to perceive the image's local structure at the pixel level. It also utilizes a degrade-and-remove module to extract degrade-insensitive features, which, along with high-level image information, guide the diffusion process. PASD can be easily integrated into existing diffusion models, such as Stable Diffusion. Experiments on realistic image super-resolution and personalized style transfer have demonstrated the effectiveness of our proposed method.