@Nick Popov Thank you for a very detailed answer! It was a pleasure to read, really :) In general, anyone who wants to create a digital copy of themselves or any other person either needs to wait for AI models to appear that can do this job, or they need to set up a live photo session where they can create Lora with the photos they receive. It is enough to recall and look at the benchmark for evaluating neural networks called "Will Smith Eating Spaghetti". That's why there will be a holiday on our street sooner or later, I think. And regarding Sydney Sweeney or Taylor Swift, I've seen several Instagram posts where AI creators create different photos with their looks. The trick of such generations is that such large models as Chat GPT Image 2.0 or Nana Banana Pro are trained on a huge amount of data, that they know perfectly well what Sydney Sweeney or Taylor Swift looks like, so they generate them with 95% accuracy, which is difficult to distinguish from a real photo.