CivArchive
    Translucent skin mechanical girl - v1.0
    NSFW
    Preview 449214
    Preview 449216
    Preview 449217
    Preview 449239
    Preview 452042
    Preview 452118
    Preview 452104
    Preview 452113

    Description

    Through translucent skin, you can see mechanical parts inside.

    How to use

    Use "translucent skin" to trigger. For more useful prompts, see my examples.

    You can use a model weight larger than 1, for example, <lora:xxx:1.3>.

    Note that you need to rename your lora file or change the lora name in prompTs to get it work.

    Negative words like "horror" should be removed, I think.

    This is a pruned lora model, you may need to update your stable diffusion webui to the latest version.

    CLIP SKIP 1

    Description

    FAQ

    Comments (15)

    gagagApr 9, 2023
    CivitAI

    What model did you use for the sample?


    lwh9346
    Author
    Apr 9, 2023

    Dark Sushi Mix 大颗寿司Mix | Stable Diffusion Checkpoint | Civitai

    But you can use whatever model you like. I've tested it on many models and it works.

    gagagApr 9, 2023

    @lwh9346 Works beautifully with Dark sushi. But on other models, it's a shame

    lwh9346
    Author
    Apr 10, 2023

    @gagag Try to use a model weigh larger than 1. I think 1.3 works for most models.

    shirokokawaiiApr 9, 2023
    CivitAI

    It is difficult to output a suit that looks similar to the sample image. :(

    lwh9346
    Author
    Apr 9, 2023

    Yes, in most models, this lora just improves the quality of bodysuit. Maybe you can try either https://civitai.com/models/24779/dark-sushi-mix-mix this model or novel AI model. I'll try to fix this later.

    lwh9346
    Author
    Apr 9, 2023

    Try this on AbyssOrangeMix2:

    <lora:translucent_skin:1.3>masterpiece, best quality, 1girl, translucent skin, purple bodysuit, space, mechanical, skeleton

    easynegative

    512*768

    Yes, this model can use a weight larger than 1 to get more effects.

    shirokokawaiiApr 9, 2023

    @lwh9346 Unfortunately, testing on AOM2 hard / Anything V3 and V5 did not give satisfactory results.

    Maybe my AI making skills are lacking.

    lwh9346
    Author
    Apr 10, 2023

    @shirokokawaii Make sure youre using the latest version of stable diffusion. I pruned the model, this may cause some problems. And remember to rename the lora file you download.

    easonchanApr 9, 2023· 2 reactions
    CivitAI

    哪个模型能跑出例图的效果啊?

    lwh9346
    Author
    Apr 9, 2023

    你可以参考一下我后面发的几个例图,使用1.3的模型权重配合提示词可以在我现有的所有模型上得到类似的效果。封面使用的是大颗寿司模型。希望你玩得开心。

    lwh9346
    Author
    Apr 9, 2023

    我发现你好像直接复制了我的prompt来进行生成,你是不是忘了把lora的名字改成你下载到的lora的文件名?

    Moonman_Apr 14, 2023· 4 reactions
    CivitAI

    to get the same results use the same lora as him.

    https://civitai.com/models/24779/dark-sushi-mix-mix

    NOVA0319Sep 8, 2023
    CivitAI

    用了大颗寿司,也调了权重,lora肯定是触发了的,别的设置照搬你的,但还是跑不出来一样的效果

    mtdnicoh759Sep 16, 2023

    clip skip 调成1