CivArchive
    Weanireal krea2 - INT8 ConvRotv1.0
    NSFW
    Preview 145754907
    Preview 145754735
    Preview 145754744
    Preview 145754804
    Preview 145754842
    Preview 145754910
    Preview 145754913
    Preview 145754915
    Preview 145754918
    Preview 145754948
    Preview 145754972
    Preview 145754990
    Preview 145755009
    Preview 145755015
    Preview 145755027
    Preview 145755033
    Preview 145755036
    Preview 145755059
    Preview 145755069
    Preview 145755098

    WeAniReal | Anime Faces with Realistic Textures

    WeAniReal is a Krea2-based merged model designed to enhance anime-style facial features while preserving realistic skin textures.

    v1.0

    Testing has shown that WeAniReal occasionally produces faces with a 2.5D appearance. If this happens, we recommend including anime-style girl in your prompt to maintain a more consistent anime-style face. WeAniReal can also generate photorealistic images when prompted with realistic-style descriptions.

    🟢System Requirements & Compatibility

    • Can run on devices with 8GB of VRAM.

    • Compatible with most Krea2 workflows.

    🟢Additional Notes

    • This release is the INT8 ConvRot quantized version.

    • The file contains only the diffusion model. A compatible text encoder and VAE are required separately.

    • Please refer to the example images on the model page for generation settings.

    🔴Disclaimer

    Results may vary depending on prompts, workflows, and generation settings. Facial, hand, or anatomical errors may occur. Operation on devices with 8GB of VRAM depends on resolution, system RAM, and offloading settings; generation speed varies by hardware. Please respect others’ privacy and likeness, and do not use this model for unlawful activities, infringement, or impersonation.

    ---------中文说明---------

    WeAniReal | 兼具写实纹理的动漫风格面部模型

    WeAniReal 是一款基于 Krea2 的融合模型，旨在增强动漫风格的面部特征，同时保留写实的皮肤纹理。

    v1.0

    测试表明，WeAniReal 在生成时偶尔会出现“2.5D”风格的面部效果。若遇到此情况，建议在提示词（prompt）中加入 anime-style girl（动漫风格女孩），以确保面部呈现出更统一的动漫风格。此外，若使用写实风格的描述词，WeAniReal 也能生成照片级写实图像。

    🟢系统要求与兼容性

    - 可在具备 8GB 显存（VRAM）的设备上运行。

    - 兼容大多数 Krea2 工作流。

    🟢补充说明

    - 本版本为 INT8 ConvRot 量化版。

    - 文件仅包含扩散模型（diffusion model）本身；需另行配置兼容的文本编码器（text encoder）和 VAE。

    - 关于生成参数设置，请参考模型页面上的示例图片。

    🔴使用须知

    本模型的生成效果会受到提示词、工作流及参数设置的影响，可能出现面部、手部或身体结构错误。8GB 显存设备的运行情况取决于分辨率、系统内存及卸载设置，生成速度因设备而异。请尊重他人的隐私与肖像，勿将本模型用于违法、侵权或冒充他人的用途

    Description

    FAQ

    Comments (5)

    qjl16114567661Oct 15, 2026· 1 reaction
    CivitAI

    COOL

    kuuhakummdOct 15, 2026· 3 reactions
    CivitAI

    不赖，好用的一批！めっちゃ使いやすい！効果が一番！Pretty good！

    AIblackProOct 15, 2026· 1 reaction
    CivitAI

    god help me finding krea2 anime lora for this

    wiehhg_37
    Author
    Oct 16, 2026

    You won’t be able to find this LoRA, man—I trained it myself and have never released it.

    catlover1937Oct 17, 2026· 18 reactions
    CivitAI

    Your Weanireal Krea2 – INT8 ConvRotv1.0 incorporates my Moody model without my permission. I have always welcomed collaboration and model merging when permission is requested, but you never asked for mine.

    I demand that you remove this version immediately and refrain from using, merging, or redistributing my work without authorization again. Please respect original creators and Civitai's model licensing and attribution requirements.

    Checkpoint
    Krea 2
    by wiehhg_37
    Download (Beta) View on CivitAI
    base model

    Details

    Downloads
    198
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/15/2026
    Updated
    10/18/2026
    Deleted
    -

    Files

    weanirealKrea2_int8Convrotv10_bf16.safetensors

    Size:
    242.05 MB
    SHA256:
    a70580f0213e67967ee9c95f05bb400e8fb08307e017a924bf3441223e023d1f

    Mirrors

    HuggingFace (328 mirrors)
    CivitAI (280 mirrors)
    TensorFiles (1 mirrors)
    ModelScope (3 mirrors)
    ModelScope CN (6 mirrors)

    weanirealKrea2_int8Convrotv10_txt.safetensors

    Size:
    4.88 GB
    SHA256:
    54bd5144df0bbc25dd6ccadfcb826b521445a1b06ae5a42570bdd2974ca87094

    Mirrors

    HuggingFace (84 mirrors)
    CivitAI (20 mirrors)
    ModelScope (1 mirrors)

    weanirealKrea2_int8Convrotv10_int8.safetensors

    Size:
    13.16 GB
    SHA256:
    340dcce4addd8907f4dfdc4372bdc03f0608de7f41e9ff8d0bf6b87f0cc06f65

    Mirrors

    CivitAI (1 mirrors)