Tutu's half-leg aurora black stockings_Zimg_V1.safetensors

162 MB · SafeTensor

Abonnement requis

<tutus-hi-silk-zit>

RunningHub Online Workflow / RunningHub 在线工作流 Ready-to-use workflow (CN): RunningHub model page (Overseas): Fan benefit: use invite code rh-v0990 to claim 1000 RH coins, plus 100 RH coins every daily login. If the coins do not appear automatically, click the avatar in the upper right and bind/activate invite code rh-v0990. 国内站工作流： RH 模型页： 粉丝福利：点击右上角头像 - 邀请码，填入 rh-v0990，可领取 1000 RH 币；每天登录还可领 100 RH 币。若没有自动到账，请在右上角绑定/激活邀请码 rh-v0990。 Created with Tutu AI Workflow Tagged with Tutu Annotator. Trained with Tutu Trainer. This LoRA is part of my Tutu AI workflow. Tutu Annotator is used for intelligent dataset annotation, prompt reverse engineering, subject-aware tag filtering, and cleaner searchable captions. For this Z-Image-Turbo experiment, the dataset was annotated with my Tutu intelligent labeling workflow and trained with Tutu Trainer. This page also connects the model back to the wider Tutu tool ecosystem: Tutu Trainer, Tutu Annotator, model workflows, and my official website. The goal is not only to release models, but to build a reusable production workflow. - Official website: - Tutu Trainer: - Tutu Annotator: - Bilibili: - YouTube: - Telegram: - TikTok: - Reddit: - GitHub: - Email: - QQ: 331506796 TuTu Hi-Silk Image for Z-Image-Turbo This is a Z-Image-Turbo experimental Hi-Silk LoRA for 8D black Aurora glossy over-knee stockings / black thighhigh stockings. It is an experimental version of my Hi-Silk direction for the Z-IMAGE workflow. The effect is already good, and the example images can be used as a reference, but there is still room for improvement. This is the first experimental version, and I may continue optimizing it later. - Base model: Z-Image-Turbo / ZImageTurbo - Type: 8D black Aurora glossy over-knee stockings / black thighhigh stockings - Trigger word: tutujiguanghaisiv1 - Recommended LoRA weight: around 0.7 - Loader: can be loaded with a normal LoRA loader - Workflow: compatible with many Z-IMAGE workflows What this model is good at - Adding 8D black Aurora glossy stocking texture in Z-Image-Turbo workflows. - Improving black over-knee / thighhigh stocking presence. - Working with ordinary LoRA loading methods. - Using Tutu Annotator labels and subject filtering to keep the dataset cleaner. - Serving as an early test for the Hi-Silk direction on Z-IMAGE. Training notes This model was made with carefully selected source material. The dataset was labeled with Tutu Annotator using the Banana 2 / Banana Pro model, then filtered with Tutu's intelligent subject filtering function. The training was completed with Tutu Trainer and the Tutu training workflow. The point of this experiment is to see how the Hi-Silk material direction transfers to Z-Image-Turbo. It is not the final form of the series yet, but it already has a useful effect and is worth testing. Recommended use - Use a normal LoRA loader. - Start with LoRA weight around 0.7. - Try different weights if the effect is too weak or too strong. - In the prompt, write that the character is wearing black over-knee stockings / black thighhigh stockings. - If you want a stronger effect, add the trigger word tutujiguanghaisiv1. - Use it inside Z-IMAGE / Z-Image-Turbo workflows. Prompt keywords Main trigger: tutujiguanghaisiv1 Useful prompt phrases: - black over-knee stockings - black thighhigh stockings - 8D black stockings - Aurora glossy black stockings - glossy black hosiery - wearing black stockings - visible stocking top - fashion portrait If the effect is weak, add tutujiguanghaisiv1 and strengthen the stocking description. If the effect is too strong, lower the LoRA weight before adding many negative prompts. Workflow note My Z-IMAGE workflow is still being supplemented and improved. If you are using my workflow resources, download and test them according to your own environment. Different Z-IMAGE workflows may need different weights and prompt details. Workflow resource: . New users can use invite code rh-v0990 if needed. FAQ Is this a finished final version? No. It is the first experimental version. The effect is already good, but I may continue optimizing it later. Do I have to use the trigger word? You can first write that the character is wearing black over-knee stockings. If you want a stronger effect, add tutujiguanghaisiv1. Which workflow should I use? It is compatible with many Z-IMAGE workflows. My own workflow resources are being expanded, so test with your local workflow and adjust weight as needed. Responsible use - 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 图图嗨丝造相 - Z-Image-Turbo 实验版 LoRA，方向是 8D 黑色极光油亮黑丝长筒袜 / 过膝袜。它是我把嗨丝方向迁移到 Z-IMAGE 工作流里的一个实验版本。 这个版本效果已经不错，可以参考例图，但还有继续提升的空间。它是第一个实验版本，后续我可能会进一步优化。 - 基础模型：Z-Image-Turbo / ZImageTurbo - 类型：8D 黑色极光油亮长筒袜 / 过膝袜 - 触发词：tutujiguanghaisiv1 - 推荐 LoRA 权重：0.7 左右 - 加载方式：正常 LoRA 加载器即可 - 工作流：兼容各种 Z-IMAGE 工作流 这个模型适合什么 - 在 Z-Image-Turbo 工作流中增加 8D 黑色极光油亮丝袜质感。 - 强化黑色长筒袜 / 过膝袜效果。 - 使用普通 LoRA 加载方式即可加载。 - 使用图图打标器和智能主体过滤，让数据集标签更干净。 - 作为嗨丝方向在 Z-IMAGE 上的早期实验版本。 训练说明 这个模型使用了比较理想的源素材。素材用图图打标器进行标注，使用 Banana 2 / Banana Pro 模型辅助打标，并用图图智能主体过滤功能过滤标签。训练使用图图训练器和图图训练工作流完成。 这个实验的重点，是测试嗨丝材质方向在 Z-Image-Turbo 里的迁移效果。它还不是这个系列的最终形态，但已经有不错效果，值得测试。 推荐用法 - 使用正常 LoRA 加载器即可。 - 推荐权重从 0.7 左右开始。 - 效果太弱或太强时，可以尝试不同权重。 - 提示词里写角色穿着黑色长筒袜 / 黑色过膝袜即可。 - 如果想进一步增强效果，可以加入触发词 tutujiguanghaisiv1。 - 放在 Z-IMAGE / Z-Image-Turbo 工作流里使用。 出图关键词 主触发词： tutujiguanghaisiv1 - black over-knee stockings - black thighhigh stockings - 8D black stockings - Aurora glossy black stockings - glossy black hosiery - wearing black stockings - visible stocking top - fashion portrait 如果效果不明显，加入 tutujiguanghaisiv1 并加强黑色长筒袜描述。如果效果太强，先降低 LoRA 权重，不要一开始就堆很多负面词。 工作流说明 我的 Z-IMAGE 工作流还在补充和完善中。如果使用我的工作流资源，请根据自己的环境下载和测试。不同 Z-IM

SHA256

1951BD138FE98E6BD4DEFB97415A38B5E8868A94ABE84F8BC96DB6B1637411CC

ID de modèle ModelTrends.ai

#23673

Taille

1536x2688

Taille

1536x2688