FIRST OF ALL: I ASK THAT IF YOU USE MY LORA; PLEASE DO NOT POST NSFW CONTENT THAT IS HARMFUL, ABUSIVE, DEGRADING, INSULTING, HUMILIATING, OR OFFENSIVE TO HUMAN DIGNITY. I DID NOT CREATE MY MODELS FOR PEOPLE TO DO THOSE THINGS. PLEASE RESPECT.
---------------
About this LoRA:
This is a new LoRa that aims to highlight the details of an anime character of any kind. It adds shadows, slightly enhances textures and beautifies the face. It also adds a slight blurred tone to the background of the image to focus attention on the character in the foreground. It should be noted that this LoRa is not perfect and this is only the first version; my plan is to improve it over time. Sincerely, I hope that if you try it, this model will be useful for you and that you will enjoy it. Thank you very much fot using this LoRa.
After the LoRa cover image, I placed some comparative images where the left side is the original image with only the base model while the right side is the same image (same base model, seed and prompt) but with the LoRa applied. Also, at the end I also attached some of the individual images. In addition, this LoRa was tested on models such as Shirataki Mix, the Animagine 3.0 and 3.1; EchoToxicXL, and AniverseXL (you are free to use any anime base model you want).
Suggested specifications:
1. Image size: 768x1152 or custom (512x768 at 2x resolution).
2. Sampling steps: 20 to 25.
3. Vae: Pastel-waifu-diffusion.vae.pt.
4. Sampling method: Euler a, DPM ++ SDE or DPM ++ 2M Karras.
5. CFG scale: 5 to 10.
6. Upscaler: R-ESRGAN 4x+ Anime6B.
7. AD detailer: optional, according to your taste and preference.
