CivArchive
    Mira Yurizaki - Dimension W - V1
    NSFW
    Preview 117189472
    Preview 117189918
    Preview 117189919
    Preview 117189917
    Preview 117189929
    Preview 117189928
    Preview 117189916
    Preview 117189926

    Mira Yurizaki Character LoRA

    A flexible character LoRA for Mira Yurizaki from the anime Dimension W, trained on 1000 curated images with diverse poses, outfits, and settings.

    What This Model Does

    This LoRA generates consistent, high-quality images of Mira Yurizaki with:

    • āœ… Accurate character features - Green hair with blue streaks, green eyes, signature headgear

    • āœ… Outfit flexibility - Works with her iconic yellow dress, bikinis, casual clothes, and more

    • āœ… Versatile settings - Beach, city streets, indoor scenes

    • āœ… Vibrant colors - Proper saturation and lighting

    Trigger Word: mira yurizaki \(dimension w\)

    Recommended LoRA Strength: 0.65-0.8
    Base Model: Illustrious-based models (tested with nlAnime_v10)
    CFG Scale: 7-8
    Sampler: DPM++ 2M Karras
    Steps: 25-35

    Example Prompts

    Basic:
    
    mira yurizaki \(dimension w\), 1girl, solo, short hair, green eyes, multicolored hair, green hair, blue hair, two-tone hair, streaked hair, smile
    Yellow Dress (Signature Look):
    
    mira yurizaki \(dimension w\), 1girl, solo, yellow dress, mechanical tail, city street, standing, outdoors, day
    Beach/Bikini:
    
    mira yurizaki \(dimension w\), 1girl, solo, bikini, beach, sitting, ocean, smile

    What It's Good At

    • Character consistency across different scenarios

    • Flexible outfit generation

    • Maintaining signature features (hair color, eyes, accessories)

    • Various poses and angles

    What It's Not Good At

    • Mechanical tail may not appear consistently (due to source material limitations)

    • Very specific canonical outfit details may vary

    Training Details

    • Images: 1000 curated, tagged images

    • Epochs: 10

    • Base Model: Illustrious

    • Tagging: WD14 auto-tagger with manual corrections


    Enjoy creating with Mira! Feel free to share your results! šŸ’š

    Description

    Recommended LoRA Strength: 0.5-0.8

    Full Usable Range: 0.2-1.0

    This LoRA is extremely flexible:

    - Lower strengths (0.2-0.4): Subtle influence, great for blending with other LoRAs

    - Mid range (0.5-0.7): Balanced, recommended for most use cases

    - Higher strengths (0.8-1.0): Maximum character accuracy, still stable

    Note: Strengths above 1.0 may cause overfitting artifacts.

    FAQ

    LORA
    Illustrious

    Details

    Downloads
    126
    Platform
    CivitAI
    Platform Status
    Available
    Created
    1/12/2026
    Updated
    4/30/2026
    Deleted
    -
    Trigger Words:
    mira yurizaki \(dimension w\),

    Files

    Mira_Yurizaki.safetensors

    Mirrors

    CivitAI (1 mirrors)