Introduction: Realistic character with a girlish touchLora model, This Lora is trained based on the majicMIX realistic V7 version base model, and additional intensive training was conducted on the facial part of the image dataset

Summon word: liqi  This is the name I gave to this Lora model (Li Qi), not including this prompting word doesn't have much impact  

Lora recommended weight suggestion (0.5-0.7)

VAE: vae-ft-mse-84000-ema

Base model suggestion: CNrealistic_MIX_V40, majicMIX realistic V7 version, Mo You humanoid, and other realistic large models have been tested CNrealistic_MIX_V40 large model output image. Strong girlish vibes are enhanced, therefore highly recommended!!

Image rendering parameters: DPM++ SDE Karras  Iteration steps: 35  Size: 512*768

Recommended enabling high-definition restoration

High-definition restoration settings: Zoom algorithm: 4X-UltraSharp Redraw magnitude 0.3-0.35     HD restoration samples : 30 Zoom factor: 1.5 times or more

Enable ADetailer facial optimization plugin, can effectively avoid facial breakdown issues

Recommended positive introductory words: (masterpiece:1.2), best quality, high resolution, (realistic:1.2), (photorealistic:1.4), (solo:1.2), (1girl:1.2), beautiful detailed eyes and face, (detailed skin texture:1), 8k, RAW photo  Note: Due to multiple intensive training on the face, when writing introductory words, please add some clothing descriptions on top of the introductory words!!

Feeling quite nervous about releasing the trained model for the first time in heart, if there's anything not good, hope the senior experts can point it out