REDZ15_DetailDaemonZ_lora_v1.1 Detail Enhancement
更新2025-12-26 17:02发布时间2025-12-26 17:02
196
0
3
REDZ15_DetailDaemonZ_lora_v1.1 Detail Enhancement - 1
头像
User_jh05wj
类型
LoRA
基础模型
Z-Image
发布时间
2025-12-26 17:02
文件签名
ecf1d25ab2a4ed3f07e3a0eca639f207dae68b231680e684ef3346657caa7063
REDZ15_DetailDaemonZ_lora_v1.1细节增强.safetensors
133.67 MB

【DetailDeamonZ-Slider Detail Control LoRA】 A "detail slider" LoRA tailor-made for ZimageTurbo / RedZ1.5, truly enabling seamless one-click toggling between "ultra-clean" and "explosive detail."

Negative Weights (-0.4 to -1.5): The further you push into negative values, the richer the detail becomes. Skin texture, fabric fibers, metallic reflections, and stray strands of hair are all maxed out—perfect for scenarios requiring top-tier precision (especially useful for fixing detail loss when stacking LoRAs on a base model).

Positive Weights (0.4 to 1.2): The higher you push it, the cleaner the image becomes. Oiliness, plastic textures, and over-sharpening are instantly smoothed out. A must-have for ultra-clean cyber aesthetics and Korean-style portraits.

Note: This LoRA does not contain all keys. If a warning about missing layers appears during loading, please ignore it; it functions normally.

Summary: Shift ZimageTurbo's detail control from "random luck" to "total freedom." You no longer need to swap through dozens of base models—simply adjust this LoRA's weight to achieve a full spectrum of detail expression, ranging from "minimalist purity" to "explosive detail."

  • Recommended Sampling Steps: 8–12 steps (loses none of Turbo's signature speed)
  • Recommended Weight Range: -1.5 to +1.5 (fine-tuning in 0.2 increments is recommended)
  • Trigger Words: None (pure style slider, unaffected by tags)

作品

暂无作品