CivArchive
    MengX Mix_2.5D - V1
    NSFW
    Preview 3083392
    Preview 3079985
    Preview 3079874
    Preview 3079917
    Preview 3080104
    Preview 3080052
    Preview 3079965
    Preview 3080074
    Preview 3080126
    Preview 3080268
    Preview 3080346
    Preview 3080362
    Preview 3080426
    Preview 3080434
    Preview 3080533
    Preview 3080536
    Preview 3080534
    Preview 3080532
    Preview 3080535
    Preview 3080537

    此模型由Fantasy_V2改良而来。画面风格更接近动漫,原因是我觉得Fantasy_V2还是偏向于写实模型,在有些方面具有很大的局限性,所以对此进行一定程度上的画风改变。

    当然我目前测试的样本也就几百张,不是很多。个人的测试效率实在是太低,而且数量太少了!

    喜欢的话可以玩玩~

    推荐设置:采样器:Euler a, 步数:30-40,分辨率512*768(请不要超过1024)

    高清修复:4x-UltraSharp15步2倍,重绘幅度:0.4

    负面词:(worst quality:1.5),(low quality:1.5),lowres,bad anatomy,bad hands,((monochrome)),((grayscale)),

    不需要额外的负面嵌入。

    推荐使用ADetailer来进行面部和手部的修复。

    AD插件的参数推荐,面部:重绘幅度0.3-0.4,手部:0.3-0.4(输入beautiful hand,可以阻止修复后出现人脸的情况)

    请根据人物的占据画面的大小来调整参数。面部特写不要开(可能会导致画面变不清晰),半身像和7分身可以设置0.3,全身可以设置0.35-0.4


    This model is improved from Fantasy_V2. The picture style is closer to animation. The reason is that I think Fantasy_V2 still prefers realistic models and has great limitations in some aspects, so I have changed the style to a certain extent.

    Of course, the samples I have tested so far are only a few hundred, not a lot. The efficiency of personal testing is really too low, and the number is too small!

    You can play it if you like~

    Recommended settings:

    sampler: Euler a, steps: 30-40, resolution 512*768 (please do not exceed 1024)

    Hires. fix Upscaler: 4x-UltraSharp, Hires steps:15, 2X, Denoising strength: 0.4

    Negative words: (worst quality:1.5),(low quality:1.5),lowres,bad anatomy,bad hands,((monochrome)),((grayscale)),

    No additional negative embedding is required.

    ADetailer is recommended for facial and hand repairs.

    Recommended parameters for the AD plug-in:

    Denoising strength: face: 0.3-0.4, hands: 0.3-0.4(Enter “beautiful hand” to prevent faces from appearing after hand repair)

    Please adjust the parameters according to the size of the screen occupied by the character. Do not turn on facial close-up (it may cause the picture to become unclear). You can set it to 0.3 for the bust and 7 avatars, and 0.35-0.4 for the whole body.

    Description

    FAQ

    Checkpoint
    SD 1.5

    Details

    Downloads
    971
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    10/21/2023
    Updated
    4/23/2026
    Deleted
    9/22/2025

    Files

    mengxMix25D_v1.safetensors

    Mirrors

    Huggingface (1 mirrors)
    CivitAI (1 mirrors)