Animavice v1.0

Vivid color. Expressive characters. Polished anime rendering.

Animavice is an Anima-based model built for bold, polished character artwork with crisp faces, expressive eyes, saturated color, luminous lighting, and detailed environments. It performs especially well on portraits, cowboy shots, full-body character illustrations, dynamic action scenes, and NSFW artwork.

The model favors a clean digital-anime finish with a lively screencap influence rather than a dark or muted mood. It can produce bright fantasy scenes, neon cityscapes, colorful outdoor settings, glossy interiors, and high-impact cover illustrations without needing an additional style LoRA.

Animavice also works well with Anima character LoRAs. The release showcase was tested across multiple characters and canon outfits, with good preservation of identity, eye color, clothing, and overall composition.

## Recommended settings

These are the settings used for the validated showcase images:

| Setting | Recommendation |

|---|---|

| Workflow | Anima |

| Steps | 28–32 — start at 30 |

| CFG | 3.5–4.5 — start at 4 |

| Sampler | euler_ancestral |

| Scheduler | normal |

| Upscaling | Generate at the base resolution, then upscale 2× |

Higher CFG is not usually necessary and can make colors, contrast, or facial features feel too aggressive.

## Quick-start recipe

Resolution: 832 × 1216
Steps: 30
CFG: 4
Sampler: er_sde
Scheduler: simple
Character LoRA: 0.8
Upscale: 2× after generation

## Recommended quality prefix

masterpiece, highres, absurdres, newest, best quality, score_9, score_8, score_7, very aesthetic, ultra-detailed

## Negative prompt

worst quality, low quality, lowres, score_1, score_2, score_3, blurry, jpeg artifacts, bad anatomy, bad hands, extra fingers, missing fingers, fused fingers, extra limbs, deformed, watermark, text, logo, artist name

Add scene-specific negatives only when needed. Very long negative prompts can unnecessarily restrict composition.

## Character LoRAs

- Start character LoRAs at 0.8.

- A useful range is 0.6–0.9.

- Lower the strength to 0.65–0.75 if the LoRA overwhelms the lighting, outfit, or facial style.

- Use the character’s trigger and canon clothing tags when identity or outfit accuracy matters.

- Avoid stacking several strong style LoRAs at once. Animavice already has a pronounced visual direction, and excessive stacking can distort eyes, faces, or color balance.

## Portrait and eye guidance

For showcase portraits, useful tags include:

head and shoulders, centered composition, looking directly at viewer, both eyes fully visible, symmetrical face, crisp pupils, natural iris texture, clean catchlights, hair swept away from eyes

Avoid conflicting eye-color tags, and reduce character-LoRA strength slightly if a LoRA produces overly stylized or inconsistent eyes.

## Full-body and action guidance

For complex poses, explicitly describe the camera and body position:

full body, dynamic low angle, feet visible, balanced stance, detailed hands, clear limb separation

Hands and crowded multi-limb poses can still vary by seed. If the overall image is strong, regenerate with a nearby seed or use a targeted detail pass instead of increasing CFG.

## Known tendencies

- The model naturally favors saturated color and strong contrast. Reduce color-intensity tags or lower CFG if the result feels oversaturated.

- Dense prompts can cause small accessories or clothing details to drift. Put the most important outfit elements earlier in the prompt.

- Strong external style LoRAs may override the model’s eyes or facial rendering.

- For elaborate backgrounds, state the location, palette, and lighting separately instead of relying on a single broad environment tag.

Animavice is designed for creators who want attractive characters, bright visual storytelling, and a polished anime finish while retaining strong compatibility with the wider Anima character-LoRA ecosystem.