File size: 11,039 Bytes
1f9dc41
 
 
 
 
cd0e726
 
 
 
 
1f9dc41
cd0e726
1f9dc41
 
 
 
cd0e726
 
9994b1a
1f9dc41
cd0e726
 
 
1ce516e
1f9dc41
cd0e726
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1f9dc41
dd7b706
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cd0e726
9ce8ff8
fe8900b
cd0e726
 
9ce8ff8
3b2b00e
cd0e726
9ce8ff8
 
cd0e726
5b02cde
73e845b
4c40600
cd0e726
 
 
73e845b
4c40600
cd0e726
9ce8ff8
cd0e726
 
 
 
3b2b00e
cd0e726
 
9ce8ff8
 
5b02cde
fb5f36d
 
9ce8ff8
 
cd0e726
fb5f36d
 
9ce8ff8
 
 
 
cd0e726
 
3b2b00e
cd0e726
 
 
 
5b02cde
5cbc3c9
 
cd0e726
 
 
5cbc3c9
 
cd0e726
 
 
9ce8ff8
07711ad
 
3b2b00e
cd0e726
07711ad
 
cd0e726
5b02cde
fb5f36d
 
cd0e726
 
 
fb5f36d
 
cd0e726
07711ad
cd0e726
 
 
 
3b2b00e
cd0e726
 
07711ad
 
5b02cde
2791200
a19c832
07711ad
 
cd0e726
2791200
e2efa06
07711ad
 
 
 
cd0e726
 
fe8900b
cd0e726
 
07711ad
3b2b00e
cd0e726
edc1db5
 
cd0e726
5b02cde
5728e72
a6e3210
cd0e726
 
 
5728e72
a6e3210
cd0e726
edc1db5
cd0e726
 
 
 
3b2b00e
cd0e726
 
edc1db5
 
5b02cde
f5579c6
4197fb2
edc1db5
 
cd0e726
f5579c6
4197fb2
edc1db5
 
 
 
cd0e726
 
3b2b00e
cd0e726
 
 
 
5b02cde
f5579c6
 
cd0e726
 
 
f5579c6
 
cd0e726
 
 
edc1db5
 
9ce8ff8
3b2b00e
cd0e726
03e9401
 
cd0e726
5b02cde
42e6272
 
cd0e726
 
 
42e6272
 
cd0e726
03e9401
cd0e726
 
 
 
3b2b00e
cd0e726
 
03e9401
 
5b02cde
42e6272
 
03e9401
 
cd0e726
42e6272
 
03e9401
 
 
 
cd0e726
 
fe83df3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
---
---
license: apache-2.0
base_model: Qwen/Qwen-VL
tags:
  - lora
  - qwen-image
  - anime
  - background
  - visual-novel
language:
  - en
pipeline_tag: text-to-image
library_name: peft
---

<div align="center">
  
### 🎨 Anime Visual Novel Backgrounds LoRA

[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Model](https://img.shields.io/badge/Model-Qwen--VL-orange.svg)](https://huggingface.co/Qwen/Qwen-VL)
[![LoRA](https://img.shields.io/badge/LoRA-Rank%208-green.svg)](#)
[![Version](https://img.shields.io/badge/Version-2.0-purple.svg)](#version-history)

---

### ⚡ Current Version Stats

| Metric | Value |
|--------|-------|
| **Training Images** | 80 anime-style backgrounds generated by seedream-4 |
| **LoRA Rank** | 8 |
| **Training Steps** | 600 |
| **Base Model** | Qwen-Image-Edit |
| **Trigger Word** | `anim3bg` |
| **Recommended Weight** | 0.7 - 0.9 |

</div>

---

## 📊 Version History

<details open>
<summary><b>Version 2.0 (Current)</b></summary>

#### Training Configuration
- **Dataset:** 80 images
- **Steps:** 600
- **LoRA Rank:** 8
- **Learning Rate:** 0.0001
- **Batch Size:** 1

</details>

<details>
<summary><b>Version 1.0</b></summary>
  
#### Training Configuration
- **Dataset:** 80 images
- **Steps:** 1000
- **LoRA Rank:** 16
- **Learning Rate:** 0.0004
- **Batch Size:** 1

</details>

---

## 💬 Trigger Word
**Required**: `anim3bg`

---

## 🚀 Usage
**Simple Prompts:**
```
anim3bg, [location]
```

**Detailed Prompts:**
```
anim3bg, [detailed location description], [time], [weather], no characters, high quality
```

---

## 📥 Download & Use

1. Download `anime_background_lora.safetensors`
2. Load in Wavespeed AI or compatible platform
3. Set weight to 0.8
4. Include `anim3bg` trigger word in prompt
5. Generate

---

<div align="center">

## 📸 Example Outputs: Before vs After Comparison (Simple Prompts)

*Testing how well the LoRA performs with minimal prompt detail*

#### **Example 1:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example01_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, library</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example01_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, library</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 2:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example002_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, beachside café</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example002_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, beachside café</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 3:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example004_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, city street at night</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example004_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, city street at night</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 4:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example003_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, forest path</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example003_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, forest path</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 5:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example005_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, train station</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example005_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, train station</code></sub><br><br>
    </td>
  </tr>
</table>

---

## 📸 Example Outputs: Before vs After Comparison (Detailed Prompts)

*Testing how well the LoRA handles complex, detailed descriptions*

#### **Example 6:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example09_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, mountain lake at sunset, dark narrow japanese alley, neon signs with japanese text, vending machines, bicycle parked on street</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example09_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, mountain lake at sunset, dark narrow japanese alley, neon signs with japanese text, vending machines, bicycle parked on street</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 7:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example10_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, empty amusement park at sunrise, blanketed in snow, Ferris wheel half-frozen, colorful lights faintly glowing through the frost</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example10_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, empty amusement park at sunrise, blanketed in snow, Ferris wheel half-frozen, colorful lights faintly glowing through the frost</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 8:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example009_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, japanese shrine festival at night, festival food and game stalls on both sides, glowing orange paper lanterns overhead, fireworks in night sky</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example009_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, japanese shrine festival at night, festival food and game stalls on both sides, glowing orange paper lanterns overhead, fireworks in night sky</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 9:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example007_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, beach with white cloud, seagulls flying, beach umbrellas and chairs in distance, white foam on turquoise water</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example007_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, beach with white cloud, seagulls flying, beach umbrellas and chairs in distance, white foam on turquoise water</code></sub><br><br>
    </td>
  </tr>
</table>

---

#### **Example 10:**

<table>
  <tr>
    <td width="50%" align="center">
      <b>❌ Without LoRA</b><br>
      <img src="example008_without_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anime-style background, mystical waterfall scene at night, cascading down rocky cliff face, rainbow visible in the mist, turquoise blue pool at base</code></sub><br><br>
    </td>
    <td width="50%" align="center">
      <b>✅ With LoRA (trigger word 'anim3bg')</b><br>
      <img src="example008_with_lora.jpeg" width="100%"><br>
      <sub>Prompt: <code>anim3bg, mystical waterfall scene at night, cascading down rocky cliff face, rainbow visible in the mist, turquoise blue pool at base</code></sub><br><br>
    </td>
  </tr>
</table>

---


# 🎨 [LoRA 名称] — Cinematic Anime Backgrounds (Outdoor)

## 📄 简介
此 LoRA 专注于生成**户外电影感(cinematic)动漫背景**:夕阳 / 金色时刻、画面感强的云与光线、宁静的日本郊区街景。  
**触发词(Trigger)**: `shinkaibg`

---

## 🧪 对比实验说明
展示相同 seed、相同 prompt 在 **未应用 LoRA / 应用 LoRA** 下的差异;以及室内提示的过拟合测试(模型偏向户外氛围)。

---

## 🔁 Same Seed — Same Prompt 比较(示例)
**Prompt 示例:** `a quiet japanese street at sunset, cinematic lighting, painterly detail, shinkaibg`

| Prompt | Without LoRA | With LoRA |
|--------|--------------|----------|
| same seed, same prompt | ![no_lora_street](images/same_seed_prompt_without_lora.png) <br> *Without LoRA* | ![with_lora_street](images/same_seed_prompt_with_lora.png) <br> *With LoRA (`shinkaibg` applied)* |

**说明:** LoRA 版本通常会加入更柔和的光线过渡、更丰富的天空细节与更强的电影感(色彩与景深)。

---

## 🏠 室内过拟合测试(Indoor Overfitting Test)
**测试说明:** 使用相同 seed,四张图展示 `simple prompt` / `detailed prompt`,并比较有无 LoRA 的输出差异。

**Prompts 示例:**
- **simple:** `a cozy living room`
- **detailed:** `a cozy japanese-style living room with warm sunlight through window, tatami floor, shinkaibg`

| Type | Without LoRA | With LoRA |
|------|--------------|----------|
| Simple prompt | ![simple_no](images/indoor_simple_without_lora.png) <br> *No LoRA* | ![simple_yes](images/indoor_simple_with_lora.png) <br> *With LoRA — 注意可能出现室外光效 / 窗外景观* |
| Detailed prompt | ![detailed_no](images/indoor_detailed_without_lora.png) <br> *No LoRA* | ![detailed_yes](images/indoor_detailed_with_lora.png) <br> *With LoRA — 需更详尽描述以避免户外化 bias* |

**观察:** 若只使用简单室内 prompt,模型往往会“套用”户外 / 日落光线(显示过拟合)。详细 prompt 能降低但不完全消除该现象。

---

## ⚙️ 推荐参数(示例)
- LoRA 权重 (strength): `0.7` - `0.9`  
- CFG Scale: `7` - `9`  
- Sampler: `DPM++ 2M Karras`(示例)  
- Resolution: `1024×1024`  

---

## 🧾 标签与数据集信息
- 数据集规模:`~[填数字]` 张户外风景图  
- 主要风格标签:`shinkaibg`, `cinematic`, `sunset`, `painterly`  
- 注意:训练集中强烈偏向户外场景,导致 indoor prompt bias(详见上方测试)

---

## 🔍 使用示例(Prompt)