PhotoBOX_Standard - Slider-LoRA Collection for ZImageTurbo
PhotoBOX_Standard - ZImageTurbo Slider-LoRA 合集
English / 中文
Overview / 简介
EN:
PhotoBOX_Standard is a collection of 7 Slider-LoRAs designed for ZImageTurbo (ZIT). Each Slider controls one specific image attribute. No trigger words needed — just load the LoRA and use (keyword: value) in your prompt (recommended range: -5 to +5, 0 is neutral).
You can stack multiple Sliders together (e.g., 4+ at once), and even with values above ±3, the results often stay coherent and can produce surprising, creative effects — that's one of the fun parts of this series!
ZH:
PhotoBOX_Standard 是专为 ZImageTurbo (ZIT) 设计的 7 个 Slider-LoRA 合集，每个 Slider 控制图像的一个特定属性。无需任何触发词，即插即用——加载 LoRA 后，直接在提示词中使用 (关键词: 值) 即可调节（推荐范围 -5 到 +5，0 为中性）。
支持 多个 Slider 叠加使用（实测可轻松叠 4 个以上），即使每个权重都在 ±3 以上，图像通常仍保持良好表现，甚至能生成一些意想不到的惊喜效果——这正是系列的乐趣之一！
All Sliders / 全部 Slider 列表
| File Name / 文件名 | Prompt Keyword / 提示词关键词 | -5 (Left) / 左端 | +5 (Right) / 右端 |
|---|---|---|---|
| Race.safetensors | (race: value) | More European / 更欧洲化 | More Asian / 更亚洲化 |
| Body.safetensors | (body: value) | Thin / 更瘦 | Fat / 更胖 |
| Aperture.safetensors | (aperture: value) | Small aperture / 小光圈 | Large aperture / 大光圈 |
| Low&High.safetensors | (quality: value) | LOW quality / 低质量 | HIGH quality / 高质量 |
| ZoomShot.safetensors | (zoom: value) | Zoom OUT / 拉远 | Zoom IN / 拉近 |
| Warm&Cool.safetensors | (color: value) | More cool / 更冷色调 | More warm / 更暖色调 |
| Dark&Light.safetensors | (exposure: value) | More dark / 更暗 | More light / 更亮 |
EN: Feel free to experiment with stacking — try high values or unexpected combinations for creative surprises!
ZH: 鼓励大胆叠加尝试——高权重或意外组合往往带来惊喜创意效果！
Usage / 使用方法
1. Diffusers (Python) / 使用 Diffusers 加载
# EN & ZH 通用示例
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"Tongyi-MAI/Z-Image-Turbo", # 官方 ZImageTurbo repo
torch_dtype=torch.bfloat16
)
# 加载多个 Slider 示例（支持任意组合）
pipe.load_lora_weights("vjleoliu/photobox-standard-zimageturbo", weight_name="Race.safetensors")
pipe.load_lora_weights("vjleoliu/photobox-standard-zimageturbo", weight_name="Body.safetensors")
pipe.load_lora_weights("vjleoliu/photobox-standard-zimageturbo", weight_name="Dark&Light.safetensors")
pipe.to("cuda")
prompt = "a beautiful portrait of a woman, (race: +4), (body: -3), (exposure: +3.5), detailed skin, realistic lighting, high quality"
image = pipe(
prompt,
num_inference_steps=8, # ZImageTurbo 推荐 8 steps
guidance_scale=3.5
).images[0]
image.save("output.png")