Type | |
Stats | 144 17 3 |
Reviews | (26) |
Published | Mar 25, 2025 |
Base Model | |
Trigger Words | chinese style |
Hash | AutoV2 8FC9CC2FF3 |
Chinese-Style Dancer v1.0
Don’t forget to submit your generated images!
Features:
• Combines 2D, 2.5D, and realistic rendering styles.
• Supports color adjustments.
• The recommended shot composition is a cowboy shot.
Example Images:
• Generated locally on an RTX 4090 (results may differ from online generation).
Recommended Models:
• PONY Base Model or models based on PONY (feel free to experiment with other base models and share results!).
---
### <Prompt Guide>
• Recommended LoRA Weight: 1~1.2
• Trigger Word: None required.
• Recommended Prompts:
Expressive, glowing, neon, starry_sky, anklet, 1girl, solo, (light smile:0.8), pale_skin, shiny_skin, black hair, chinese style architecture, darkkhaki, foreshortening, night_sky, fantasy
• Negative Prompts:
score_6, score_5, score_4
• Note: PONY models require minimal quality modifiers.
• Feel free to experiment with additional elements or common tags!
---
### <Generation Parameters>
• Sampling Method: Euler a
• Sampling Steps: 25+
• Aspect Ratios: 1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9 (other ratios welcome).
• CFG Scale: 5–7
• Resolution:
• Use 800x1200 with AD Direct Output or apply 1.1~1.5x HD upscaling.
• For higher precision:
◦ HD Upscale Denoising Strength: 0.35–0.5
◦ Upscaler: 4x-UltraSharp
---
### <Notes>
• Common Issues: Hands, body, or limb distortions may occur—reroll generations if needed!
• Local Use Recommended: For best results, download models locally.
• Model Names: Ignore example image LORA names; they’re for archiving only.
• Minimum Resolution: 800x1200 (lower resolutions may cause facial distortions).
• ADetailer Fix: Enable ADetailer and select face_yolov8n
for facial repairs.
---
Let me know if you need further tweaks! 😊