tutu's HiSilk (5D Polka Dot Black Pantyhose)/图图的嗨丝（5D波卡圆点黑色连裤袜丝袜）/チュチュのハイシルク（5D水玉黒パンスト）

zhaotutuzhaotutu
424
2.9K
性感femaleclothespantyhose服装girls
v2.0v1.0
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

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.

Polka-dot hosiery is harder than plain hosiery because the model must learn both the black nylon material and the repeating dot pattern without spreading the dots into clothes, skin, shoes, or the background. This is exactly why cleaner captions and repeated tests matter.

Tutu AI Workflow - Tutu Trainer and Tutu Annotator

Tutu's HiSilk 5D Polka Dot Black Pantyhose LoRA

This is a HiSilk LoRA for 5D black polka-dot pantyhose. It focuses on ultra-thin black pantyhose texture with a polka-dot pattern. The current recommended version is v2.0, made with Difference Extraction Method 3.0 and real-shot material.

The earlier polka-dot versions were difficult: the dots were hard to trigger and the pollution was heavy. After many, many attempts, I rebuilt the model. v2.0 is much better than before: easier to trigger, less contaminated, and more practical in normal workflows.

  • Brand / product reference: Moumi / limited polka-dot series
  • Type: polka-dot black pantyhose
  • Thickness: 5D
  • Color: black
  • Base model: SD 1.5
  • v2 trigger: (tutututu), pantyhose
  • Optional stronger dot prompt: tutututu, polka dot pantyhose
  • Recommended v2 weight: 0.4 to 1.0

What this model is good at

  • Creating 5D black pantyhose with a visible polka-dot pattern.
  • Adding a more realistic black nylon surface instead of flat painted dots.
  • Reducing the dot-pattern pollution that can spread to other parts of the image.
  • Working well with fashion, portrait, 2.5D, and realistic workflows.
  • Pairing especially well with black footwear, loafers, and leather boots.

Why polka-dot pantyhose needed its own LoRA

Polka-dot hosiery is not just black pantyhose plus a few dots. The pattern has scale, spacing, density, and material behavior. If the model learns it poorly, the dots may become dirty texture, random spots on the skin, pattern pollution on clothing, or a noisy background.

This is why I rebuilt it after the earlier versions did not satisfy me. The goal of v2.0 is to keep the dots easier to trigger while controlling their influence. The model should add a fashionable polka-dot pantyhose look, not contaminate the whole picture.

Recommended settings

  • Make sure the LoRA is loaded correctly.
  • Recommended CLIP skip / CLIP stop layers: 2.
  • Recommended weight: 0.4 to 1.0. Try several values depending on the base model and desired dot strength.
  • Start CFG Scale around 7.
  • Do not overload negative prompts, or the LoRA effect may be suppressed.
  • Use black footwear, loafers, or leather boots when you want a cleaner matching fashion look.

Prompt keywords

Stable v2 trigger:

(tutututu), pantyhose

Stronger dot trigger:

tutututu, polka dot pantyhose

The stronger dot trigger can make the polka dots more obvious, but the word polka dot is also more likely to affect other parts of the image. If that happens, try adding BREAK or lower the LoRA weight.

Useful prompt phrases:

  • polka dot pantyhose
  • black polka dot pantyhose
  • 5D black pantyhose
  • ultra-thin black pantyhose
  • black nylon pantyhose
  • small polka dots
  • fashion pantyhose
  • black footwear
  • loafers
  • leather boots

Training and production notes

v2.0 uses Difference Extraction Method 3.0 and real-shot production. The model was rebuilt after many attempts because the earlier polka-dot versions were too polluted and difficult to trigger. The new version uses a cleaner trigger strategy and the unified tutututu trigger family for easier use across my later LoRAs.

Difference extraction is useful here because the model must focus on the pantyhose material and the dot pattern, not the entire source photo. For patterned hosiery, this distinction is extremely important.

Difference Extraction Method 3.0

Normal LoRA training can learn too much from the source images: lighting, background, base-model bias, clothing, shoes, or unrelated texture. Difference extraction tries to extract the useful difference and remove more of those contaminants.

For my own material, I also optimized the process with automated model exploration and Euclidean-distance-based difference analysis between LoRA weights. In simple terms, LoRA weights can be treated as points in high-dimensional space. Comparing distances between them helps guide fusion and selection decisions. This is not a generic one-click trick; it is adjusted around the quirks of my material.

Recommended models and resources

These were part of the original recommendation notes and remain useful as direction references:

  • tutu's face / 图图的脸模 / 花ひよこ 顔モデル
  • Muggle scene LoRAs: Japanese restaurant, escalator perspective, nightclub/bar, passenger-seat girlfriend, French window, fitting room selfie
  • Moyou artificial human
  • Moyou FACE_realistic
  • MoYou 2.5D_MIX
  • LEOSAM's Instant photo / Polaroid
  • LEOSAM's FilmGirl / VELVIA
  • LEOSAM's Clothing +/- Adjuster

Version history

v2.0 - current recommended version. Rebuilt after many attempts. Easier to trigger, less contaminated, made with Difference Extraction Method 3.0 and real-shot material. Use (tutututu), pantyhose; optionally add polka dot pantyhose for stronger dots.

v1.0. Original version using tutubd and black pantyhose. It worked best on 2.5D base models, then realistic models, and could work on anime models with careful adjustment. It used 1024 x 1024 + bf16 material processing, but the later v2.0 is easier and cleaner.

FAQ

Why do the dots spread to clothes or skin?

The word polka dot can affect the whole prompt, not only the pantyhose. Use the stable trigger first, add polka dot pantyhose only when needed, try BREAK, and lower the weight if necessary.

Why are the dots too weak?

Increase the LoRA weight slowly, add polka dot pantyhose, and use a base model that understands patterned hosiery. Some base models simplify the pattern.

What shoes look good with it?

Black footwear, loafers, and leather boots usually match the feeling better.

Can I use it commercially?

You are responsible for what you generate. If you use this model for commercial work or custom cooperation, please contact me first.

Statement

  • 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 5D 波卡圆点黑色连裤袜 LoRA。它的目标是超薄黑色连裤袜质感，同时带有清楚、好看的波点图案。当前推荐版本是 v2.0，使用差异提取法 3.0 + 实拍制作。

之前做的波点总是做不好，污染很大。经过很多很多次尝试，才做出这个新版本。v2.0 比之前好太多了：更容易触发，污染更少，也更适合正常工作流使用。

  • 品牌 / 产品参考：某觅限量款波卡圆点
  • 类型：波点黑色连裤袜
  • 厚度：5D
  • 颜色：black
  • 基础模型：SD 1.5
  • v2 触发：(tutututu), pantyhose
  • 更强波点触发：tutututu, polka dot pantyhose
  • v2 推荐权重：0.4 到 1.0

这个模型适合什么

  • 生成 5D 黑色连裤袜和清晰波点图案。
  • 表现更真实的黑色尼龙丝袜表面，而不是简单画几个点。
  • 减少波点污染到衣服、皮肤、背景的问题。
  • 适合时装、人像、2.5D、写实等工作流。
  • 搭配 black footwear、乐福鞋、皮靴通常更好看。

为什么波点需要单独做

波点连裤袜不是“黑丝 + 几个点”这么简单。波点有大小、间距、密度，也要贴合丝袜材质。如果模型学得不好，波点会变成脏纹理、皮肤上的随机斑点、衣服上的污染，或者背景噪点。

所以 v2.0 才要重做。它的目标是让波点更容易触发，同时控制影响范围。它应该生成时髦的波点连裤袜，而不是污染整张图。

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：0.4 到 1.0，根据底模和你想要的波点强度多试。
  • CFG Scale 可以从 7 开始。
  • 负面提示词不要堆太多，否则会影响 LoRA 效果。
  • 鞋子建议尝试 black footwear、乐福鞋、皮靴等，整体更搭。

出图关键词

稳定触发：(tutututu), pantyhose

更强波点触发：tutututu, polka dot pantyhose

polka dot 会让波点更明显，但也更容易影响图像其他地方。如果出现污染，可以加 BREAK，或降低 LoRA 权重。

  • polka dot pantyhose
  • black polka dot pantyhose
  • 5D black pantyhose
  • ultra-thin black pantyhose
  • black nylon pantyhose
  • small polka dots
  • fashion pantyhose
  • black footwear
  • loafers
  • leather boots

制作说明

v2.0 使用差异提取法 3.0 和实拍素材制作。因为之前版本的波点比较难触发，污染也重，所以这个版本经过了很多次尝试和重做。新的版本统一到 tutututu 触发体系，后续使用会更方便。

差异提取法在这个模型里很重要，因为模型要学的是丝袜材质和波点图案，而不是整张素材照片。对于图案类丝袜 LoRA，这个区别非常关键。

差异提取法 3.0

普通 LoRA 直接拿图训练，很容易把灯光、背景、底模偏好、衣服、鞋子等不相关内容一起学进去。差异提取法的目标，是尽量提取真正有用的差异，减少这些污染。

我针对自己的素材做了自动化模型探索和基于欧氏距离的权重差异分析。简单说，就是把 LoRA 权重看成高维空间里的点，通过比较距离来辅助融合和选择。这不是通用的一键技巧，而是围绕我自己的素材特性做的取舍。

推荐模型和资源

这些是原文里的推荐方向，仍然可以作为搭配参考：

  • tutu's face / 图图的脸模 / 花ひよこ 顔モデル
  • 麻瓜场景系列：日料店、扶梯视角、夜店酒吧、副驾驶女友、法式落地窗、真实试衣间自拍
  • 墨幽人造人
  • 墨幽 FACE_realistic
  • MoYou 2.5D_MIX
  • LEOSAM's Instant photo / Polaroid
  • LEOSAM's FilmGirl / VELVIA
  • LEOSAM's Clothing +/- Adjuster

版本说明

v2.0：当前推荐版本。 经过很多次尝试后重做，更容易触发，污染更少，使用差异提取法 3.0 + 实拍素材。使用 (tutututu), pantyhose；需要更强波点时可加入 polka dot pantyhose。

v1.0： 原始版本，触发词为 tutubd 和 black pantyhose。在 2.5D 底模下效果最好，真人模型其次，二次元也能出但需要微调。使用 1024 x 1024 + bf16 处理，但 v2.0 更容易、更干净。

常见问题

为什么波点会跑到衣服或皮肤上？

polka dot 这个词可能影响整个提示词，不只影响丝袜。先用稳定触发，需要时再加 polka dot pantyhose，可以尝试 BREAK，必要时降低权重。

为什么波点不明显？

缓慢提高权重，加入 polka dot pantyhose，并选择更理解图案丝袜的底模。有些底模会简化图案。

搭配什么鞋更好看？

black footwear、乐福鞋、皮靴通常更适合这个模型。

声明

  • 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 5D polka dot black pantyhose LoRA です。薄い黒パンストの質感と、水玉模様を表現するためのモデルです。おすすめは v2.0 で、Difference Extraction Method 3.0 と実写素材を使って作成しました。

  • Type: polka-dot black pantyhose
  • Thickness: 5D
  • Color: black
  • Stable trigger: (tutututu), pantyhose
  • Stronger dot prompt: tutututu, polka dot pantyhose
  • Recommended weight: 0.4 to 1.0

polka dot を入れると模様は強くなりますが、服や肌など他の部分に影響する場合があります。その場合は BREAK を試すか、LoRA ウェイトを下げてください。black footwear、loafers、leather boots との組み合わせがおすすめです。

此模型生成的图像