Choosing the Right Model: ControlNet or LoRa for Face-Centric Image Generation?

I’m seeking guidance on selecting the best model for a specific image generation task that focuses on the face. I plan to upload 10 images of my face and train a model to generate new images based on custom prompts while ensuring my face remains the central element. For instance, if I prompt “me in a blue suit,” the generated image should feature myself in a blue suit.

Considering the emphasis on facial features, would ControlNet or LoRa be more suitable for this application? I’d greatly appreciate any advice or insights on which model or method would excel in preserving and enhancing facial details.

Thank you for your assistance!