anim3bg-visual-novel-background-v2

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

---

⚡ Current Version Stats

</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> _{Prompt: <code>anime-style background, library</code>}<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> _{Prompt: <code>anim3bg, library</code>}<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> _{Prompt: <code>anime-style background, beachside café</code>}<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> _{Prompt: <code>anim3bg, beachside café</code>}<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> _{Prompt: <code>anime-style background, city street at night</code>}<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> _{Prompt: <code>anim3bg, city street at night</code>}<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> _{Prompt: <code>anime-style background, forest path</code>}<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> _{Prompt: <code>anim3bg, forest path</code>}<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> _{Prompt: <code>anime-style background, train station</code>}<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> _{Prompt: <code>anim3bg, train station</code>}<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> _{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>}<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> _{Prompt: <code>anim3bg, mountain lake at sunset, dark narrow japanese alley, neon signs with japanese text, vending machines, bicycle parked on street</code>}<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> _{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>}<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> _{Prompt: <code>anim3bg, empty amusement park at sunrise, blanketed in snow, Ferris wheel half-frozen, colorful lights faintly glowing through the frost</code>}<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> _{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>}<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> _{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>}<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> _{Prompt: <code>anime-style background, beach with white cloud, seagulls flying, beach umbrellas and chairs in distance, white foam on turquoise water</code>}<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> _{Prompt: <code>anim3bg, beach with white cloud, seagulls flying, beach umbrellas and chairs in distance, white foam on turquoise water</code>}<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> _{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>}<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> _{Prompt: <code>anim3bg, mystical waterfall scene at night, cascading down rocky cliff face, rainbow visible in the mist, turquoise blue pool at base</code>}<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

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

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）