CivArchive
    Kasumi - v1.0
    NSFW
    Preview 140134991
    Preview 140134973
    Preview 140134971
    Preview 140134972
    Preview 140134974
    Preview 140134980
    Preview 140134970
    Preview 140134979
    Preview 140134986
    Preview 140134984
    Preview 140134987
    Preview 140134989
    Preview 140134992
    Preview 140134994
    Preview 140134995
    Preview 140134997
    Preview 140134998
    Preview 140135001
    Preview 140136244
    Preview 140136306

    Kasumi — V1 | Realistic Character LoRA

    Kasumi V1 is a realistic character LoRA inspired by Kasumi from Dead or Alive, created to reproduce her recognizable identity while giving you freedom to place her in completely different outfits, environments, poses, and photographic styles.

    The goal with this LoRA was not simply to reproduce one familiar costume or one specific look.

    The goal was to create a flexible, realistic interpretation of Kasumi that can move naturally between martial arts, fashion, lifestyle photography, cinematic scenes, casual environments, and more — while still remaining recognizably Kasumi.

    Put her in a dojo.

    Dress her for the office.

    Take her to the beach.

    Drop her into a neon city at night.

    Give her casual clothing, evening wear, athletic gear, streetwear, tactical clothing, or traditional Japanese outfits.

    The character should remain consistent while the rest of the image changes around her.


    Version

    Current release: V1

    This is the first public version of the Kasumi LoRA.

    V1 establishes the character identity, realistic appearance, body type, hair, facial features, and general versatility of the model.

    Future versions may continue refining character consistency, flexibility, pose handling, clothing variation, and photographic realism.


    Trigger

    Use:

    kreakaz

    For the strongest character consistency, I recommend beginning your prompts with:

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail.

    From there, simply continue with the outfit, pose, environment, lighting, composition, camera direction, and photographic style you want.


    Character Appearance

    Kasumi V1 interprets the character as a realistic adult woman with:

    • Long auburn hair

    • High ponytail

    • Auburn eyes

    • Soft but recognizable facial features

    • Curvy adult proportions

    • Strong identity consistency across different scenes

    • A natural photographic appearance rather than an anime-rendered look

    The recommended prompt prefix helps anchor these features while leaving the rest of the prompt open for experimentation.


    Prompting

    Kasumi works especially well with natural-language photographic prompts.

    You do not need enormous tag lists or complicated prompt structures.

    Start with:

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail.

    Then describe the image you want normally.


    Example — Martial Arts

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail. Wearing a burgundy athletic top, fitted grey training leggings and white sneakers, practicing martial arts inside a traditional wooden Japanese dojo, captured mid-kick, focused expression, sunlight streaming through the windows, dust particles suspended in the light, realistic sports photography, natural anatomy, detailed skin and fabric texture, photorealistic.


    Example — Professional

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail. Wearing a perfectly tailored black business suit over a crisp white blouse and black heels, walking confidently through a modern glass corporate lobby carrying a structured handbag, natural mid-stride pose, soft morning sunlight, polished stone reflections, sophisticated editorial photography, realistic skin texture, full-body composition, photorealistic.


    Example — Beach

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail. Wearing a flowing pale blue summer dress, walking barefoot along a tropical beach at sunset while carrying her sandals in one hand, looking back toward the camera with a relaxed smile, ocean waves rolling onto the sand, warm golden backlight illuminating her ponytail, realistic lifestyle photography, natural skin texture, soft cinematic depth of field, photorealistic.


    Wardrobe Freedom

    You do not need to keep Kasumi in her traditional outfit.

    Experimenting with wardrobe is encouraged.

    Try:

    • Athletic wear

    • Business clothing

    • Dresses

    • Knitwear

    • Streetwear

    • Jeans

    • Jackets

    • Evening fashion

    • Tactical clothing

    • Traditional Japanese clothing

    • Summer outfits

    • Casual home clothing

    • Original costume designs

    The intention is for the LoRA to preserve Kasumi's identity while allowing the prompt to control the wardrobe.


    Photography Tips

    Photographic language works extremely well with Kasumi V1.

    Try adding terms such as:

    35mm photography

    50mm lens

    85mm portrait lens

    shallow depth of field

    natural window light

    golden hour

    cinematic lighting

    editorial photography

    street photography

    lifestyle photography

    realistic skin texture

    natural proportions

    subtle film grain

    photorealistic

    Camera direction can also make a substantial difference.

    For example:

    full-body shot

    medium shot

    three-quarter portrait

    low camera angle

    environmental portrait

    captured mid-motion

    walking toward the camera

    looking over her shoulder

    dynamic action pose


    Dynamic Poses

    Don't feel limited to static portrait poses.

    Kasumi V1 works well with movement and action.

    Try:

    • Martial arts

    • Walking

    • Running

    • Turning

    • Crouching

    • Leaning

    • Sitting

    • Training

    • Mid-kick poses

    • Natural candid movement

    • Interacting with objects or environments

    Her long ponytail also works especially well in motion and can become a strong visual element within the composition.


    The sample gallery intentionally includes very different environments, clothing styles, lighting conditions, poses, and camera distances.

    That variety is deliberate.

    A character LoRA should not only be able to reproduce one carefully selected portrait.

    The more interesting test is whether the character remains recognizable when almost everything else about the image changes.

    That is one of the main things Kasumi V1 is designed to demonstrate.


    Disclaimer

    This is a fan-made character LoRA and is not an official Dead or Alive or Koei Tecmo product.

    Kasumi and Dead or Alive are properties of their respective rights holders.

    This LoRA is intended as a creative character-generation model for AI image generation.


    Model: Kasumi
    Version: V1
    Trigger: kreakaz

    Recommended character anchor:

    kreakaz. kasumi, 26 year old curvy woman with auburn eyes and long auburn hair in high pony-tail.

    Have fun experimenting with her — Kasumi doesn't have to stay in the dojo.

    Description

    Release Notes — V1

    • Initial public release

    • Realistic interpretation of Kasumi

    • Strong auburn hair and high-ponytail identity

    • Auburn eye consistency

    • Curvy adult body type

    • Good wardrobe flexibility

    • Strong lifestyle and fashion rendering

    • Good full-body generation

    • Martial arts and action pose capability

    • Natural photographic rendering

    • Works across indoor and outdoor environments

    • Designed for significant scene and clothing variation

    Kasumi V1 is the foundation release and an ongoing project.

    FAQ

    LORA
    Krea 2

    Details

    Downloads
    126
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/19/2026
    Updated
    8/26/2026
    Deleted
    -

    Files

    Kasumi-Krea2-V1.safetensors

    Mirrors

    CivitAI (1 mirrors)