👉日本语版
Core Technique for Dramatically Improving Portrait Quality
——Combining Quality Tokens with Negative Prompts
I. Quality Tokens ：Quality tokens fall into two groups.
①General-purpose quality tokens
Common ones include: masterpiece, best quality, ultra-detailed / highly detailed, realistic / photorealistic, etc.
②Model-specific quality tokens
Certain models provide dedicated labels or score-type tokens to produce a particular style or higher-rated images. Example: score_9 (exclusive tag for the Pony model).
II. Negative Prompts：Negative prompts tell the AI “what should not appear” in the image.
Their three key purposes are:
1.Remove flaws
2.Eliminate distracting elements
3.Refine style and composition
III. Comparison Demo (Pony Realism Model)
Open Advanced Settings and check “Negative Tags.”
1) No Quality Tokens / No Negative Prompts
Prompt：①
②
→ Generated Images：
①
②
2) With Quality Tokens / No Negative Prompts (Sharper Quality)
Prompt：
①
②
→ Generated Images：
①
②
3) With Quality Tokens and Negative Prompts (High Quality + Clean Composition)
Prompt：①
②
→ Generated Images：
①
②
