Images with an AR higher than 2:3 use HighRes Fix. Images vary between 70% (.7) - 100% (1) LoRA strength. Base Model (used for training and for example images): https://civitai.com/models/66/anything-v3
This LoRA is aimed to better enhance anime portraiture with a focus on males (can do females, but isn't trained on them). Still a WIP, so you might get niche results. Let me know your thoughts.
Any LoRA strength is fine. Higher strengths might give more masculine/male results, whereas lower strengths might allow more femininity and follow the prompt more. It's really all up to you boo.
Training images generated by me.