In my previous and earlier works, many of the faces ended up looking similar, and I created nearly ten failed LoRAs along the way. This time, it almost became another failed attempt, but when I saw the video I used as the cover, I felt it came close to my ideal and decided to publish it.I think this prompt might have worked well. Please use the following prompt as a reference 🙇‍♀️  A young Japanese woman with a small V-shaped face, slim delicate jawline, tiny chin, enlarged doll-like double-lidded eyes, long fluttering eyelashes, small nose, thin lips, porcelain-like skinSince the results trained with Wan2.2 turned out slightly different from the original training images, I decided to release a Wan2.1 version that stays closer to the trained face. I tested this LoRA with both the High-Noise and Low-Noise settings in Wan2.2, but I have not tested it with Wan2.1. Still, I believe it should work fine.Of course, this is just my personal impression, and some people may still find the faces look similar. I kindly ask for your understanding on that point.On Civitai, I am constantly inspired by the many wonderful works I see, and I enjoy creating my own LoRAs by adding my own arrangements and ideas. Until now, many of my works had the same “big-eyed droopy look,” but recently I have been challenging myself to create faces that look cute, beautiful, and elegant, like idols, actresses, or hostesses.When it comes to prompt creation, I always rely on ChatGPT for support. While being mindful of copyright and portrait rights, I carefully select safe LoRAs and keep challenging myself to try new expressions. There are many failures along the way, but the process of trial and error with ChatGPT as my partner, together with everyone’s reactions, always encourages me.It is truly an honor and a joy that some of the creators I deeply respect have chosen to follow me. Even though my works may sometimes look similar, I sincerely hope you will continue to support me. 🙏🙇‍♀️💕Thank you very much for reading this long post.

Files with SHA256: c3e30a19c0900db27069ddda65a7bd0c2c12246b728dac2b44333509d80c072e