I discovered the optimal training parameters in SDXL 1.0. The generated LORA now resembles a real person in real life.

Train based on SDXL 1.0

Dataset using over 100 carefully selected high quality images

Please follow my Facebook AI_Imaging

Join my discord server https://discord.gg/pJ6GGcgAGV for free discussion

My Patreon https://www.patreon.com/AI_Imaging/membership for exclusive LORAs

*************************

Best weight is from 0.8 to 1

Steps: 40-60, Sampler: DPM++ 2M Karras, CFG scale: 7

Model: majicmixRealistic_v6, Denoising strength: 0.3-0.4, Clip skip: 2, Hires upscale: 2, Hires steps: 16-30, Hires upscaler: 4x-UltraSharp

******************************************************

In our Telegram channel - AI Imaging: The new ERA you'll find exciting content related to AI and imaging.

We also have a Telegram group Free Group- AI Imaging where you can engage in discussions and connect with like-minded individuals.

Additionally, on our YouTube channel - AI Imaging Studio, you can explore a wide range of AI-related content.