This model, focusing on "high-precision figure drawing+film level light and shadow texture", achieves a natural balance between anime illustration and realistic rendering, and can quickly output character illustration, game CG and virtual image materials that meet commercial visual standards.
Style adaptation: compatible with Japanese illustration, ACG vertical drawing, 3D rendering texture, and supporting multiple visual expressions from aesthetic anime to realistic.
Detail accuracy: The restoration of features, hair texture, clothing wrinkles, and skin texture is extremely high, and it can generate rich layers of light, shadow, and scene atmosphere.
Basic Generation Parameters
| Parameter Item | Recommended Value | Description |
|---|---|---|
| Sampler | DPM++ 2M Karras / Euler a | Balances generation speed and detail presentation |
| Sampling Steps | 20-30 steps (25 steps recommended) | More steps deliver richer details; 25 steps strike a balance between efficiency and quality |
| Resolution | 1024×1536 / 1536×1024 | Ideal for vertical character illustrations and artwork, supports 2K output |
| Aspect Ratio | 2:3 (character illustration) / 1:1 (avatar/card design) | Matches visual proportion of reference images |
| Prompt Weight | Positive: 1.0-1.2 / Negative: -1.0~-1.5 | Highlight superior features and avoid low-quality elements |
High-Resolution Fix
Enable it, upscale ratio: 1.5-2.0, denoising strength: 0.2-0.4, to enhance detail sharpness
Face Restoration
Turn on CodeFormer with intensity 0.6-0.8 for refined details and accuracy
Contrast Enhancement
Apply Latent Upscale plus sharpen filter to boost picture clarity and visual vividness
Seed Value
Fixed seed ensures identical effect reproduction; random seed creates diverse visual outputs