CivArchive
    Makima: The Control Devil | Anima LoRA - v1.0
    NSFW
    Preview 143544042
    Preview 143544081
    Preview 143544082
    Preview 143544087
    Preview 143544093
    Preview 143544095
    Preview 143544119
    Preview 143544131
    Preview 143544137
    Preview 143544143
    Preview 143544151
    Preview 143544158
    Preview 143544164
    Preview 143544166
    Preview 143544169
    Preview 143544175
    Preview 143544184

    Makima: The Control Devil | Anima LoRA

    Makima character LoRA for the Anima model family, trained for strong character identity while remaining flexible across outfits, poses, environments, lighting conditions, and camera angles.

    The goal of this LoRA was not simply to reproduce Makima's standard Public Safety appearance. It was built to maintain Makima's character identity across a wide variety of concepts, allowing you to place her into completely different situations while retaining the features that make her recognizable.

    The included showcase images were generated across a wide range of conditions, including close-up portraits, full-body compositions, dynamic poses, casual clothing, formal dresses, her classic Public Safety outfit, fantasy and sci-fi concepts, indoor and outdoor environments, daylight, nighttime, cinematic lighting, and Control Devil-inspired scenes.


    Trigger Word

    makima_csm

    Place the trigger near the beginning of your prompt.

    Example:

    makima_csm, 1girl, solo, adult woman, ...


    Recommended Strength: 0.7

    The showcase images for this LoRA were generated at 0.7 strength, which provided a strong balance between Makima's character identity and the flexibility of the underlying checkpoint.

    Higher strengths may increase the LoRA's influence, but they can also begin to overpower the checkpoint's rendering style, clothing, anatomy, or other prompt instructions.

    For results similar to the showcase images, start at 0.7 and adjust only if necessary for your particular checkpoint.


    Character Features

    The LoRA was trained to reproduce Makima's recognizable characteristics, including:

    Strawberry-blonde / auburn-red hair • blunt bangs • center bang • long braided ponytail • golden-yellow eyes with red concentric rings • mature feminine appearance • recognizable Makima facial features

    The LoRA is not dependent on her canonical outfit.

    Makima can be prompted in completely different clothing while maintaining her character identity, including casual outfits, dresses, fantasy clothing, futuristic designs, and more.


    For stronger character consistency, keep the trigger word and important character features near the beginning of your prompt.

    A good starting point is:

    makima_csm, 1girl, solo, adult woman, mature woman, strawberry blonde hair, auburn red hair, blunt bangs, single thin center bang, long thick braided ponytail, golden yellow eyes, clearly defined red concentric rings inside irises

    After establishing Makima's identity, describe the clothing, pose, expression, environment, camera angle, lighting, and art style you want.

    For example:

    makima_csm, 1girl, solo, adult woman, mature woman, strawberry blonde hair, auburn red hair, blunt bangs, single thin center bang, long thick braided ponytail, golden yellow eyes, clearly defined red concentric rings inside irises, black fitted turtleneck, blue jeans, walking through a modern city, natural daylight, cinematic composition, premium 2.5D anime

    The LoRA was specifically tested with significant changes to wardrobe and environment rather than only Makima's canonical appearance.


    Makima's Eyes

    Makima's eyes are one of her most recognizable features, but they're also a detail that can become difficult for the underlying checkpoint to resolve when her face occupies only a small portion of the image.

    The eye wording that produced the most consistent results during testing was kept relatively simple:

    golden yellow eyes, clearly defined red concentric rings inside irises, detailed eyes

    One important lesson from testing: more eye prompting is not necessarily better.

    Overloading the prompt and negative prompt with multiple competing eye-color instructions can actually reduce consistency. If you're having problems with her eyes, start with the simple wording above before adding additional eye tags.

    Close-up and medium-distance compositions naturally provide the model with more resolution for rendering the concentric rings.

    Face detailing or an additional enhancement pass can also help when working with full-body compositions.


    Suggested Negative Prompt

    This is a general starting point rather than a requirement:

    low quality, worst quality, lowres, blurry, flat 2d, flat lighting, photorealistic, child, young girl, loli, childish face, short hair, bob cut, missing braid, multiple braids, malformed braid, malformed eyes, cross-eyed, asymmetrical eyes, missing concentric rings, bad anatomy, bad hands, malformed hands, extra fingers, missing fingers, fused fingers, six fingers, extra arms, extra legs, missing limbs, duplicate person, multiple girls, text, watermark, logo

    I recommend adjusting your negative prompt to suit the checkpoint, composition, and style you're using rather than treating this as mandatory.


    What This LoRA Was Built For

    The main priority was Makima herself rather than locking the LoRA into one outfit, pose, or composition.

    It was tested for character consistency while changing:

    Wardrobe • Pose • Expression • Camera angle • Camera distance • Environment • Lighting • Art direction

    The final showcase contains 20 curated generations covering a broad range of situations.

    These include straightforward identity-focused portraits, difficult full-body compositions, dynamic movement, seated anatomy, casual clothing, formal clothing, Makima's traditional Public Safety appearance, daylight environments, nighttime environments, cinematic scenes, fantasy concepts, and supernatural Control Devil imagery.

    The goal of the showcase is to demonstrate what the LoRA can do rather than showing twenty variations of essentially the same Makima image.


    Model Compatibility

    This LoRA was created for the Anima model family.

    Because this is a character LoRA rather than a complete checkpoint, the checkpoint you pair it with will still have a major influence on:

    Rendering style • anatomy • hands • lighting • textures • composition • prompt adherence

    Results can therefore vary between Anima-family checkpoints.

    If you're trying the LoRA for the first time, I recommend starting at 0.7 strength before making other adjustments.


    Final Notes

    This LoRA was extensively tested and curated before release, with the goal of producing a Makima who remains recognizable even when taken far outside her standard outfit and environment.

    Start with:

    Trigger: makima_csm
    LoRA Strength: 0.7

    Then build whatever version of Makima you want around it.

    Have fun experimenting with her — Makima doesn't have to wear the same white shirt and tie forever.

    Description

    Initial release of the Makima character LoRA. Trained for strong character consistency across different outfits, poses, environments, lighting conditions, and compositions.

    LORA
    Anima

    Details

    Downloads
    38
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/22/2026
    Updated
    9/23/2026
    Deleted
    -
    Trigger Words:
    makima_csm

    Files

    makima_csm_anima-v1.safetensors

    Mirrors