RF-Inversion is a generative AI system that efficiently inverts reference style images and applies semantic edits based on text prompts without requiring textual descriptions of the source images. It uses rectified stochastic differential equations for image generation, enabling applications like content-preserving stylization and controlled image manipulation. The model is available as a Gradio demo, integrated into the Hugging Face diffusers library, and supported in ComfyUI.