Upload v1.0/README.md with huggingface_hub
Browse files- v1.0/README.md +170 -0
v1.0/README.md
ADDED
|
@@ -0,0 +1,170 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
tags:
|
| 5 |
+
- art
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# Qwen Edit Omni Cosplay By Aldniki
|
| 9 |
+
|
| 10 |
+
**Creator**: [aldniki217](https://civitai.com/user/aldniki217)
|
| 11 |
+
**Type**: LORA
|
| 12 |
+
**Base Model**: Qwen
|
| 13 |
+
**Version**: v1.0
|
| 14 |
+
**Trigger Words**: `生成一张真实的照片,一个模特穿着图中的衣服和配饰进行穿搭拍摄,保持发型和刘海和表情和服装和配饰不变`
|
| 15 |
+
|
| 16 |
+
**Civitai Model ID**: 1940046
|
| 17 |
+
**Civitai Version ID**: 2195734
|
| 18 |
+
|
| 19 |
+
**Stats (at time of fetch for this version)**:
|
| 20 |
+
* Downloads: 209
|
| 21 |
+
* Rating: 0 (0 ratings)
|
| 22 |
+
* Favorites: N/A
|
| 23 |
+
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
## 📄 Description (Parent Model)
|
| 27 |
+
|
| 28 |
+
You can use the online workflow below. If the results are good, you can then try setting it up and running it locally.
|
| 29 |
+
👉 Click Here For Online WorkFlow
|
| 30 |
+
As a fan benefit, you’ll receive 1,000 points upon registration, plus 100 points for daily logins. Enjoy a smooth experience with an RTX 4090 and 48GB of VRAM.
|
| 31 |
+
Model Description
|
| 32 |
+
This LoRA is a full upgrade of my Reality Transform series, achieving a
|
| 33 |
+
more realistic transformation
|
| 34 |
+
from anime to live-action photos. It better preserves facial expressions and adds greater diversity to facial features, overcoming the previous LoRA's limitation of extremely similar face characteristics.
|
| 35 |
+
This LoRA enables the following features (or enhances the corresponding functions of the Qwen Image Edit base model):
|
| 36 |
+
Transforming anime characters into realistic photos
|
| 37 |
+
Keeping the same outfit, but replacing the model
|
| 38 |
+
Generating model photos from plain white background clothing images
|
| 39 |
+
Numerous background or scene replacement options (see the prompt section for details)
|
| 40 |
+
Recommended Settings
|
| 41 |
+
Sampler
|
| 42 |
+
: Euler
|
| 43 |
+
LoRA Weight
|
| 44 |
+
: 1
|
| 45 |
+
CFG
|
| 46 |
+
: 1.0 (Using the
|
| 47 |
+
Qwen-Image-Lightning-8steps-V1.1
|
| 48 |
+
, acceleration LoRA weight 1.0)
|
| 49 |
+
Steps
|
| 50 |
+
: 8
|
| 51 |
+
Recommended to use high-resolution upscale:
|
| 52 |
+
1.5× scale. Denoising set to 0.45. Beta scheduler is recommended, with 8 steps, CFG 1.0, using the
|
| 53 |
+
Qwen-Image-Lightning-8steps-V1.1
|
| 54 |
+
, acceleration LoRA weight 0.75.
|
| 55 |
+
Recommended Prompt (In Chinese)
|
| 56 |
+
You can also write your prompts in English!
|
| 57 |
+
The following description may be lengthy, so it is still strongly recommended to check out my online workflow, which includes all the prompt options mentioned below.
|
| 58 |
+
👉 Click Here For Online WorkFlow
|
| 59 |
+
Prompts can consist of the following components:
|
| 60 |
+
[Trigger words]
|
| 61 |
+
+
|
| 62 |
+
[Whether to lock the original pose and angle]
|
| 63 |
+
+
|
| 64 |
+
[Whether to complete the original character's body]
|
| 65 |
+
+
|
| 66 |
+
[Scene options (feel free to use your imagination)]
|
| 67 |
+
↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓
|
| 68 |
+
[Trigger Words]
|
| 69 |
+
:
|
| 70 |
+
生成一张真实的照片,一个模特穿着图中的衣服和配饰进行穿搭拍摄,保持发型和刘海和表情和服装和配饰不变,
|
| 71 |
+
[Pose and Angle Lock Options]:
|
| 72 |
+
1.保持姿势和拍摄角度不变,
|
| 73 |
+
(lock pose and angle)
|
| 74 |
+
2.适当地调整姿势和拍摄角度,
|
| 75 |
+
(adjust pose and angle)
|
| 76 |
+
[Full Body Completion Options]:
|
| 77 |
+
补全模特的整个躯体,镜头拉远,展示从头到鞋子的所有穿搭,
|
| 78 |
+
(Complete the model's entire body, zoom out the camera)
|
| 79 |
+
[Scene Options] (Leave this one blank to keep the original background
|
| 80 |
+
)
|
| 81 |
+
1.背景替换成日本comiket漫展展厅室内,模特摆出优雅的拍照姿势,背后有其他穿着各种各样动漫服装的cosplayer在走动,没有留意到模特,
|
| 82 |
+
(Japanese comiket convention indoor)
|
| 83 |
+
2.背景替换成日本comiket漫展户外,模特摆出优雅的拍照姿势,
|
| 84 |
+
(Japanese comiket convention outdoor)
|
| 85 |
+
3.背景替换成中国chinajoy漫展展厅室内,模特摆出优雅的拍照姿势,
|
| 86 |
+
(Chinese chinajoy convention indoor)
|
| 87 |
+
4.背景替换成咖啡店,咖啡店里没有人,模特盘腿坐在沙发上,摆出优雅的拍照姿势,视角换成俯拍,
|
| 88 |
+
(sitting on a sofa in a cafeteria)
|
| 89 |
+
5.背景替换成街边商铺门口,模特摆出优雅的拍照姿势,视角转变成正面照,
|
| 90 |
+
(outside the entrance of a street-side shop)
|
| 91 |
+
6.模特坐在满是人的麦当劳店里,手里拿着一个巨无霸汉堡正打算吃掉,桌子上的餐盘里放着很多炸鸡翅和麻辣鸡腿堡和麦辣鸡块,
|
| 92 |
+
(in a McDonald restaurant)
|
| 93 |
+
7.模特站在站立的观众人群之中,双手举过头顶欢呼,灯光氛围很强,
|
| 94 |
+
(in a livehouse)
|
| 95 |
+
8.模特坐在战斗机的的驾驶舱内,正在驾驶战斗机在天空中飞行,驾驶舱被一个玻璃罩盖住,改变模特的姿势为双手握住战斗机的方形方向舵,拉远镜头拍摄到飞机的全貌,
|
| 96 |
+
(flying a fighter jet)
|
| 97 |
+
9.模特站在一个冲浪板上,在大海上进行冲浪,改变动作为双手张开向两侧平举,双腿微曲,摆出冲浪的姿势,
|
| 98 |
+
(surfing)
|
| 99 |
+
10.模特正站在月球表面,远处可以看到地球悬挂在宇宙中,模特单手插腰,另一只手举着一面写有艺术字体的“COSPLAY FOREVER”的旗帜,模特的鞋子踩在灰白色的月尘上,她的身后有一辆巨大的月球车。
|
| 100 |
+
(on the surface of the moon)
|
| 101 |
+
...
|
| 102 |
+
...
|
| 103 |
+
Many more scenes to explore!
|
| 104 |
+
模型介绍
|
| 105 |
+
这个 LoRA 是我 Reality Transform 系列的全面升级版本,能够将���次元角色
|
| 106 |
+
更真实地转化为写实照片
|
| 107 |
+
,同时更好地保留人物表情,并使面部特征更加多样化,解决了之前 LoRA 面部特征过于相似的问题。
|
| 108 |
+
此 LoRA 可以实现以下功能(或增强 Qwen Image Edit 底模的对应功能):
|
| 109 |
+
将二次元角色转化为写实照片
|
| 110 |
+
指定同一套服装,但是替换模特
|
| 111 |
+
从白底图生成模特照片
|
| 112 |
+
多种背景或场景替换玩法(详见提示词部分)
|
| 113 |
+
推荐设置
|
| 114 |
+
采样器
|
| 115 |
+
: Euler
|
| 116 |
+
LoRA 权重
|
| 117 |
+
: 1
|
| 118 |
+
CFG
|
| 119 |
+
: 1.0 (使用
|
| 120 |
+
Qwen-Image-Lightning-8steps-V1.1
|
| 121 |
+
, 加速LoRA权重1.0)
|
| 122 |
+
步数
|
| 123 |
+
: 8
|
| 124 |
+
推荐使用高清放大
|
| 125 |
+
: 1.5倍. 降噪 0.45. Beta 调度器, 8步, CFG 1.0, 使用
|
| 126 |
+
Qwen-Image-Lightning-8steps-V1.1
|
| 127 |
+
, 加速LoRA权重0.75
|
| 128 |
+
推荐提示词
|
| 129 |
+
以下的描述说明可能字数较多,还是强烈推荐去使用我的配套的在线工作流,里面内置了以下所有提示词选项。
|
| 130 |
+
👉 在线工作流
|
| 131 |
+
提示词可以由以下几个部分组成:
|
| 132 |
+
[触发词] + [是否锁定原图的姿势和角度] + [是否补全原图的人物身体] + [场景选项(请发挥你的想象)]
|
| 133 |
+
↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓
|
| 134 |
+
[触发词]
|
| 135 |
+
:
|
| 136 |
+
生成一张真实的照片,一个模特穿着图中的衣服和配饰进行穿搭拍摄,保持发型和刘海和表情和服装和配饰不变,
|
| 137 |
+
[锁定姿势和角度选项]:
|
| 138 |
+
1.保持姿势和拍摄角度不变,
|
| 139 |
+
2.适当地调整姿势和拍摄角度,
|
| 140 |
+
[补全身体选项]:
|
| 141 |
+
补全模特的整个躯体,镜头拉远,展示从头到鞋子的所有穿搭,
|
| 142 |
+
[场景选项] (这一项不写可以保留原背景)
|
| 143 |
+
1.背景替换成日本comiket漫展展厅室内,模特摆出优雅的拍照姿势,背后有其他穿着各种各样动漫服装的cosplayer在走动,没有留意到模特,
|
| 144 |
+
2.背景替换成日本comiket漫展户外,模特摆出优雅的拍照姿势,
|
| 145 |
+
3.背景替换成中国chinajoy漫展展厅室内,模特摆出优雅的拍照姿势,
|
| 146 |
+
4.背景替换成咖啡店,咖啡店里没有人,模特盘腿坐在沙发上,摆出优雅的拍照姿势,视角换成俯拍,
|
| 147 |
+
5.背景替换成街边商铺门口,模特摆出优雅的拍照姿势,视角转变成正面照,
|
| 148 |
+
6.模特坐在满是人的麦当劳店里,手里拿着一个巨无霸汉堡正打算吃掉,桌子上的餐盘里放着很多炸鸡翅和麻辣鸡腿堡和麦辣鸡块,
|
| 149 |
+
7.模特站在站立的观众人群之中,双手举过头顶欢呼,灯光氛围很强,
|
| 150 |
+
8.模特坐在战斗机的的驾驶舱内,正在驾驶战斗机在天空中飞行,驾驶舱被一个玻璃罩盖住,改变模特的姿势为双手握住战斗机的方形方向舵,拉远镜头拍摄到飞机的全貌,
|
| 151 |
+
9.模特站在一个冲浪板上,在大海上进行冲浪,改变动作为双手张开向两侧平举,双腿微曲,摆出冲浪的姿势,
|
| 152 |
+
10.模特正站在月球表面,远处可以看到地球悬挂在宇宙中,模特单手插腰,另一只手举着一面写有艺术字体的“COSPLAY FOREVER”的旗帜,模特的鞋子踩在灰白色的月尘上,她的身后有一辆巨大的月球车。
|
| 153 |
+
...
|
| 154 |
+
...
|
| 155 |
+
更多场景可以探索!
|
| 156 |
+
|
| 157 |
+
## Civitai Links
|
| 158 |
+
|
| 159 |
+
* **[🔗 View This Version on Civitai →](https://civitai.com/models/1940046?modelVersionId=2195734)**
|
| 160 |
+
* [View Full Model Page →](https://civitai.com/models/1940046)
|
| 161 |
+
* [View Creator Profile →](https://civitai.com/user/aldniki217)
|
| 162 |
+
|
| 163 |
+
---
|
| 164 |
+
|
| 165 |
+
## File Information
|
| 166 |
+
|
| 167 |
+
* **Filename**: `aldniki_qwen_omni_cosplay_v01.safetensors`
|
| 168 |
+
* **Size**: 281.47 MB
|
| 169 |
+
* **Hash (AutoV2)**: `EF20F4BE3F`
|
| 170 |
+
* **Hash (SHA256)**: `EF20F4BE3F412D3C57B0A7EBD08766C06E8836699E28B487B273FE8C23E21859`
|