tutu’s HiSilk (LMERExxxxM Thor series 20D shiny complexion T-crotchless stockings pantyhose)/图图的嗨丝(X觅雷神系列20D油亮肤色T裆无腰丝袜连裤袜)

zhaotutuzhaotutu
304
2.6K
服装
v1.0
model-image-0
model-image-1
model-image-2
model-image-3
model-image-4
model-image-5
model-image-6
model-image-7
model-image-8
model-image-9
model-image-10
model-image-11
model-image-12
model-image-13
model-image-14
model-image-15
model-image-16

详情

下载文件 (1)

模型描述

Created with Tutu AI Workflow

Trained with Tutu Trainer. Tagged with Tutu Annotator.

This LoRA is part of my Tutu AI workflow. Tutu Trainer is used for LoRA training, model resource management, parameter presets, and training checks. Tutu Annotator is used for dataset annotation, prompt reverse engineering, English-friendly tags, and cleaner searchable captions.

For HiSilk models, accurate training and tagging matter a lot. The LoRA should improve hosiery texture, thickness, shine, and color, while avoiding pollution to clothing, shoes, body shape, background, and the rest of the image.

Tutu AI Workflow - Tutu Trainer and Tutu Annotator

Tutu HiSilk: 20D Shiny Complexion Pantyhose

This is a HiSilk LoRA for 20D shiny complexion / skin-color pantyhose, based on the LMERExxxxM Thor series direction. The target is a SAND-tone, glossy, thin nylon pantyhose look with T-crotchless / no-waist styling.

Compared with ordinary pantyhose prompts, this model focuses on the actual texture: thin 20D coverage, skin-color shine, realistic nylon surface, and a cleaner full-leg hosiery effect. It is better suited for realistic and 2.5D models. It usually does not perform as well on strong anime models.

  • Type: 20D shiny complexion pantyhose / skin-color HiSilk LoRA
  • Color: SAND / skin tone
  • Style: T-crotchless, no-waist pantyhose direction
  • Base model: SD 1.5
  • Trigger words: tutuhsls, pantyhose, white pantyhose
  • Recommended weight: 0.5 to 0.9, start around 0.6
  • Recommended CLIP skip: 2
  • CFG Scale: start around 7

What this model is good at

  • Improving the texture of skin-color shiny pantyhose.
  • Restoring a thin 20D nylon look instead of generic leg color or opaque tights.
  • Keeping the hosiery effect realistic on real / 2.5D models.
  • Reducing pollution to shoes, clothing, body shape, background, and overall composition.
  • Letting users fine-tune the result through trigger-word weights.

Why I remade this direction

Starting from the one-piece pantyhose series, I kept optimizing the production method. Friends who used my older LoRAs may know that the earlier models still had some pollution. They did not usually damage the body or background, but they could still affect content outside the stockings, such as clothing or shoes.

This skin-color version was reset again because I wanted a cleaner result. The texture is much better than before, and the most important improvement is pollution control. Under normal conditions, it should not pollute unrelated elements in the image.

The difference extraction method I use now is no longer the earliest method published online. It has been improved through my own testing. The goal is simple: affect the pantyhose texture as much as possible, and affect everything else as little as possible.

Why another pantyhose LoRA?

There are already many similar LoRAs, but I found two common problems. First, many models pollute the image: sometimes the clothing changes, sometimes the shoes change, sometimes the body shape changes. Second, many models do not subdivide hosiery thickness, color, and style clearly, so everything becomes mixed together.

That is why I made this model and why I continue building the HiSilk series. Different thicknesses, colors, shine levels, and structures should not be treated as the same thing. I am quite confident about this one: for skin-color shiny pantyhose, this should be one of the most useful versions I have made.

Recommended settings

  • Make sure the LoRA is loaded correctly.
  • Recommended CLIP skip / CLIP stop layers: 2.
  • Recommended weight: 0.5 to 0.9. Start around 0.6.
  • Start CFG Scale around 7.
  • Do not add too many negative prompts, or the LoRA effect may be weakened.
  • This model is better for realistic and 2.5D base models. Strong anime models are not recommended.

Prompt keywords and control tips

Main trigger group:

tutuhsls, pantyhose, white pantyhose

The trained word white pantyhose is part of the control set for this model. Although the target color is skin/SAND, increasing the influence of white pantyhose can make the hosiery look cleaner and brighter. Increasing tutuhsls strengthens the skin-color shiny texture.

  • skin color pantyhose
  • shiny pantyhose
  • 20D pantyhose
  • thin nylon pantyhose
  • glossy nylon texture
  • sheer pantyhose
  • realistic legs
  • fashion portrait
  • studio lighting

If the result is too pale or too white, reduce the weight of white pantyhose or lower the LoRA weight. If the skin-color shine is weak, strengthen tutuhsls and describe the pantyhose texture more clearly.

Training notes

All source photos were taken from real high-definition camera material to keep the effect realistic. The model was carefully made with my improved difference extraction workflow. The point is not only to make the legs look shiny, but to make the hosiery itself more convincing while keeping the rest of the image stable.

If you encounter problems, please leave me a message. I will continue optimizing the production method as I make more HiSilk models.

FAQ

Why does it not work well on anime models?

This LoRA is based on real-shot material and focuses on realistic hosiery texture. Realistic and 2.5D models are more suitable.

Why does the result become too white?

The trained word white pantyhose can brighten the hosiery. Lower its weight or reduce the LoRA weight if the SAND / skin-color feeling is lost.

Why does it affect clothing or shoes?

Lower the LoRA weight first. Also describe the clothing, shoes, and scene clearly so the model does not use hosiery texture to fill nearby areas.

Why can I not get exactly the same image as the examples?

Different base models, samplers, LoRAs, plugins, and settings will produce different results. Please use the examples as a direction and create your own best version.

Responsible use

  • Use this LoRA responsibly and only with appropriate adult / fashion contexts when creating mature hosiery images.
  • You are fully responsible for any creative work made with this model.
  • Do not use this model to create or share illegal, harmful, malicious, defamatory, fraudulent, or political content.
  • If you use this model commercially, please inform me. Thank you.

中文说明

这是图图嗨丝系列里的 20D 油亮肤色连裤袜 LoRA，方向是 X觅雷神系列、SAND 肤色、油亮、T 裆无腰丝袜连裤袜。它的重点不是简单让腿变亮，而是改善肤色连裤袜的真实质感。

这个模型更适合真人大模型和 2.5D 模型，在强二次元大模型上的表现通常不佳。它的目标是薄 20D 覆盖、肤色油亮、尼龙质感，以及尽量不污染鞋子、衣服、身材、背景和整体构图。

  • 类型：20D 油亮肤色连裤袜 / 肤色嗨丝 LoRA
  • 颜色：SAND / 肤色
  • 样式：T 裆无腰连裤袜方向
  • 基础模型：SD 1.5
  • 触发词：tutuhsls, pantyhose, white pantyhose
  • 推荐权重：0.5 到 0.9，建议从 0.6 开始
  • CLIP 终止层：2
  • CFG Scale：从 7 左右开始

这个模型适合什么

  • 改善肤色油亮连裤袜的质感。
  • 还原 20D 薄透尼龙效果，而不是普通腿色或厚袜。
  • 在真人 / 2.5D 模型里表现更自然。
  • 尽量减少对鞋子、衣服、身材、背景和构图的污染。
  • 可以通过触发词权重微调颜色和质感。

为什么要重做这个方向

从连体丝袜系列开始，我一直在优化制作方法。用过我早期 LoRA 的朋友应该知道，之前的模型还是有一定污染。虽然通常不会影响身材、背景等大内容，但还是可能对丝袜以外的衣服、鞋子等部分产生影响。

所以我又重置了这个肤色款。这次质感比之前好很多，更重要的是污染控制好了很多。正常情况下，它应该不会对其他画面元素产生明显污染。

现在我使用的差异提取法已经不是最早网上公开的那种方法，而是我自己不断测试后改良的版本。目标很明确：尽量只影响丝袜质感，尽量不影响其他内容。

为什么已经有这么多类似 LoRA 还要做

因为我发现很多现有 LoRA 有两个问题。第一，会污染画面：要么衣服变了，要么鞋子变了，要么身材被影响。第二，没有把丝袜厚度、颜色、样式分清楚，整体是比较混乱的状态。

所以我继续做嗨丝系列。不同厚度、颜色、光泽、结构，本来就不应该混成一种东西。自信点，这个肤色丝袜 LoRA 应该是我做过的肤色丝袜方向里非常好用的版本。

推荐参数

  • 确保正确加载 LoRA。
  • CLIP 终止层数建议设置为 2。
  • 推荐权重：0.5 到 0.9，建议从 0.6 开始。
  • CFG Scale 可以从 7 开始尝试。
  • 负面提示词不要堆太多，否则会影响 LoRA 效果。
  • 更推荐真人、写实、2.5D 模型，不太推荐强二次元大模型。

出图关键词和控制方法

主触发词组：

tutuhsls, pantyhose, white pantyhose

white pantyhose 是这个模型的控制词之一。虽然目标是肤色 / SAND，但加强 white pantyhose 会让丝袜更白、更干净；加强 tutuhsls 会增强肤色油亮质感。

  • skin color pantyhose
  • shiny pantyhose
  • 20D pantyhose
  • thin nylon pantyhose
  • glossy nylon texture
  • sheer pantyhose
  • realistic legs
  • fashion portrait
  • studio lighting

如果结果太白，就降低 white pantyhose 或 LoRA 权重。如果肤色油亮感不够，就加强 tutuhsls，并更明确描述丝袜质感。

制作说明

这个模型全部采用真人高清相机实拍素材制作，尽量保证效果真实。制作时使用了我改良后的差异提取流程，目的不是把整张图变亮，而是让丝袜本身更有说服力，同时保持其他画面元素稳定。

如果使用中遇到问题，可以给我留言。我会随着嗨丝系列继续优化制作方法。

常见问题

为什么二次元模型表现不好？

这个 LoRA 基于真实实拍素材，更注重真实丝袜纹理，所以真人和 2.5D 模型更适合。

为什么结果太白？

white pantyhose 会让丝袜更白、更干净。如果失去 SAND 肤色感，请降低它的权重或降低 LoRA 权重。

为什么衣服或鞋子被影响？

先降低 LoRA 权重，同时把衣服、鞋子和场景描述清楚，避免模型把丝袜质感扩散到附近区域。

为什么我出不了和例图一样的图片？

每个人使用的底模、采样器、LoRA、插件和参数都不一样，请把例图当作方向，不要强求完全一致。

声明

  • 涉及成熟丝袜内容时，请只用于合适的成年 / 时装语境。
  • 你需要对使用本模型生成的任何作品承担全部责任。
  • 不要使用本模型制作或传播非法、有害、恶意、诽谤、诈骗或政治用途内容。
  • 如果将本模型用于商业用途或定制合作，请提前联系我，谢谢。

日本語説明

これは Tutu HiSilk シリーズの 20D shiny complexion / skin-color pantyhose LoRA です。SAND 系の肌色、薄い 20D のナイロン感、光沢、T-crotchless / no-waist pantyhose 方向を表現するためのモデルです。

  • Trigger: tutuhsls, pantyhose, white pantyhose
  • Weight: 0.5 to 0.9, start around 0.6
  • CLIP skip: 2
  • CFG Scale: start around 7

リアル系または 2.5D モデルに向いています。強いアニメ系モデルでは効果が弱くなることがあります。white pantyhose を強めると白く清潔な方向に寄り、tutuhsls を強めると肌色の光沢感が強くなります。服や靴に影響する場合は LoRA ウェイトを下げてください。

此模型生成的图像