Hypothetically, yes, it's very possible, but I don't like using real people in base images, I prefer crafting things from scratch. That being said, I can create a low-level SD model called a "LoRA" that I'd use on top of whatever other model I'm using, and it's purpose would be to allow me to recreate whatever that LoRA is based on.

For example, if you give me a bunch of high quality (at least 1024 x 1024 pixel) images of a particular person, the LoRA will make it so that every image I generate with that LoRA will look like that person. You can do that same thing with objects (cars, clothes, flowers, etc.), styles ("in the style of RossDraws, Vincent van Gogh, Disney Pixar, etc.), compositions ("background full of lightning", "circular japanese watercolor", "BIG dick anal penetration from behind"), or even just make one to add more details or correct for common issues like fixing hands or eyes or whatever.

Like I said though, my specialty is pure prompting - just using a model and prompt to create completely unique images from scratch. But if you intend to have me illustrate the same character repeatedly, it will be necessary to make a LoRA to do so.