Update README.md
Browse files
README.md
CHANGED
|
@@ -26,6 +26,35 @@ LoRA fine-tuned on Qwen-Image for generating consistent anime-style backgrounds
|
|
| 26 |
- **Dataset Size**: 80 anime backgrounds
|
| 27 |
- **Specialization**: Visual novel backgrounds with time/weather control
|
| 28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
## Trigger Word
|
| 30 |
|
| 31 |
**Required**: `anim3bg`
|
|
|
|
| 26 |
- **Dataset Size**: 80 anime backgrounds
|
| 27 |
- **Specialization**: Visual novel backgrounds with time/weather control
|
| 28 |
|
| 29 |
+
## Example Outputs
|
| 30 |
+
|
| 31 |
+
### 🎨 Before vs After - See the Difference
|
| 32 |
+
|
| 33 |
+
This LoRA dramatically improves consistency, style, and detail in anime background generation.
|
| 34 |
+
|
| 35 |
+
---
|
| 36 |
+
|
| 37 |
+
<table>
|
| 38 |
+
<tr>
|
| 39 |
+
<th colspan="2" align="center">Mountain Lake - Clear Sky</th>
|
| 40 |
+
</tr>
|
| 41 |
+
<tr>
|
| 42 |
+
<td width="50%" align="center">
|
| 43 |
+
<b>❌ Base Model (No LoRA)</b><br><br>
|
| 44 |
+
<img src="./examples/comparison/mountain_lake_without_lora.png" width="100%"><br>
|
| 45 |
+
<sub>Base model output — lacks visual novel tone and depth</sub>
|
| 46 |
+
</td>
|
| 47 |
+
<td width="50%" align="center">
|
| 48 |
+
<b>✅ With LoRA (0.8 weight)</b><br><br>
|
| 49 |
+
<img src="./examples/comparison/mountain_lake_with_lora.png" width="100%"><br>
|
| 50 |
+
<sub>“anim3bg” style — consistent anime lighting, smooth reflections, vivid colors</sub>
|
| 51 |
+
</td>
|
| 52 |
+
</tr>
|
| 53 |
+
</table>
|
| 54 |
+
|
| 55 |
+
**Prompt:** `anim3bg, mountain lake with wooden dock, small rowboat tied to pier, forested mountains reflecting in calm water, clear blue sky with white clouds, ripples in water, no people`
|
| 56 |
+
|
| 57 |
+
|
| 58 |
## Trigger Word
|
| 59 |
|
| 60 |
**Required**: `anim3bg`
|