This is a LoRA trained on Stability AI's Stable Diffusion XL Base 1.0, specifically for the character Shiho. It's designed to generate "shiho" in a wide variety of poses and situations.
Key Features:
Base Model: Stable Diffusion XL Base 1.0
Training Method: LoRA / LyCORIS / Native Fine-Tuning (DreamBooth method)
Training Data: 108,396 total steps, equivalent to approximately 42,333 images.
Recommended Resolution: Variable bucket resolutions focusing on 512x512.
Optimizer: 8-bit AdamW optimizer, bf16 mixed precision.
Dropout: 0.2
How to Use: Include "Shiho" in your prompt. Adjust the LoRA weight between 0.7 and 1.0 as a guideline, and experiment to find the best results for your desired output.
Content Note: The training data includes images of various poses and specific situations.
