Update on v0.5/更新版本v0.5:Rearrange the tag priority to improve the training output, now training different outfits in separate files to improve the uniformity of the single image. The generalization has greatly increased and can be used with other lora. Trained with Anythingv5, works fine with AnythingV5 and setusMix, please try for yourself for other models.将tag的优先度进行了调整以提升训练输出的效果，并将不同的衣服分开训练以提高单个文件的统一度。目前版本的泛化程度大大提高，可以与其他lora更好的同时使用。基于AnythingV5训练，对AnythingV5以及setusMix使用的效果良好，您可以自由尝试和其他的模型组合。Suggest weights/建议权重: 0.3~0.5Trigger Words/触发词：hiseki erio, Erio_casual/Erio_yoshiwara/Erio_show (Depending on the model/第二个触发词基于模型会有所不同）Suggested Prompts/建议咒语: Casual Clothes （经典服）:solo, 1girl, hiseki erio, Erio_casual, green_eyes, red_hair, single thighhigh, garter straps, uneven legwear, collarbone, breasts, cleavage, hair_ornament, black_tank_top, denim_shorts, off_shoulder, black thighhighs, navel, stomach, striped, striped_thighhighs, short_hair, green jacketJapanese Clothes (和服）：solo, 1girl, hiseki erio, Erio_yoshiwara, green_eyes, red_hair, collarbone, breasts, japanese_clothes, sarashi, bondages, cleavage, hair_ornament, hair_flower, flower, earrings, skirt, floral print, off_shoulder, black thighhighs, green skirt, breast bondage, obi, sash, kimono, sarashi,zettai ryouiki, yellow flower,Show Clothes （演出服）：solo, 1girl, hiseki erio, Erio_show, green_eyes,red_hair, collarbone, breasts, cleavage, gloves, black gloves, fingerless gloves, green_skirt, shorts, navel, thighhighs, high_ponytails, ponytail, hair ornament, gold_striped thighhighs, lace-trimmed legwear, white belt, beltUpdate on v0.4 / 更新版本v0.4:Trained with better tagger and improve the details of different clothes, setting weight at 0.5~0.7 would be good. Check the trigger words from each clothes on the sample images.使用了更好的标签工具并提升了每一件衣服的细节，0.5~0.7的权重较为适合，具体的每一个触发词可以看对应衣服的范例。This lora model is trained based on the original authors of the <a target="_blank" rel="ugc" href="https://huggingface.co/WarriorMama777/OrangeMixs">OrangeMix</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/4437/abyssorangemix2-sfw">AbyssOrangeMix</a>, by downloading and using this model, you are agreeing to the <a target="_blank" rel="ugc" href="https://huggingface.co/spaces/CompVis/stable-diffusion-license">CreativeML OpenRAIL-M</a> and any further licenses applied from the OrangeMix. Please read the full license <a target="_blank" rel="ugc" href="https://huggingface.co/spaces/CompVis/stable-diffusion-license">here</a>.This work is only trained with anime figures, and no real person data is included during the training process. Please respect the people and local laws that apply when generating or publishing your work based on this model on the internet.这个lora模型是基于<a target="_blank" rel="ugc" href="https://huggingface.co/WarriorMama777/OrangeMixs">OrangeMix </a>中的<a target="_blank" rel="ugc" href="https://civarchive.com/models/4437/abyssorangemix2-sfw">AbyssOrangeMix</a> 进行训练的。下载并使用此模型，意味着您认同 <a target="_blank" rel="ugc" href="https://huggingface.co/spaces/CompVis/stable-diffusion-license">CreativeML OpenRAIL-M</a> 协议，以及其他在OrangeMix作者追加的使用协议。完整的协议，请阅读<a target="_blank" rel="ugc" href="https://huggingface.co/spaces/CompVis/stable-diffusion-license">此处</a>。这个模型的训练仅使用了二次元图像，没有任何真人相关的数据在训练中被使用。使用此模型生成或发表图像时，请尊重他人，并遵守相关的法律法规。(中文在下方）English brief:The model is based on the avatars of the Vup (Vtuber) on the <a target="_blank" rel="ugc" href="https://space.bilibili.com/407106379/">Bilibili</a>/<a target="_blank" rel="ugc" href="https://www.youtube.com/@hiseki_eriochannel3611">Youtube</a>, Hiseki Erio. Six L2D clothes are trained and embedded in the current version. Please feel free to follow/subscribe her account if you like this model.The main model training is based on the <a target="_blank" rel="ugc" href="https://civarchive.com/models/4437/abyssorangemix2-sfw">AOM2</a>, and it should work fine with <a target="_blank" rel="ugc" href="https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main">SD1.5</a>, <a target="_blank" rel="ugc" href="https://huggingface.co/rim0/dreambox-mix">Dreambox</a>, AOM2, and AOM3 of any type.The suggested baking weight for the Lora is 0.5~0.8.The keywords of training come with one main keyword “Erio” and six other keywords depending on the selection of the L2D style you want to produce: “casual_clothes”, “japanese_clothes”, “show_clothes”, “loli_clothes”, “leather_clothes” and “china_dress” (for V0.2). The keywords may change in different versions. Some of the example prompts are provided in the figures. I did not modify the prompts in the figures so that might be some difference in the model versions or keywords in some figures. But it should be fine as a starting hint :)Please enjoy.中文简介：这个模型是基于在<a target="_blank" rel="ugc" href="https://space.bilibili.com/407106379/">Bilibili</a>和<a target="_blank" rel="ugc" href="https://www.youtube.com/@hiseki_eriochannel3611">Youtube</a>活跃的Vup/Vtuber绯赤艾莉欧的二次元形象生成的。6件L2D的服装被整合在当前的版本中。如果您喜欢这个模型，欢迎关注/订阅她的账号。模型的训练的底模基于<a target="_blank" rel="ugc" href="https://civarchive.com/models/4437/abyssorangemix2-sfw">AOM2</a>，实际测试中，应当支持<a target="_blank" rel="ugc" href="https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main">SD1.5</a>, <a target="_blank" rel="ugc" href="https://huggingface.co/rim0/dreambox-mix">Dreambox</a>, AOM2和AOM3相关底模的图片生成。建议的模型lora权重为0.5~0.8。主要的关键词为：”Erio”。可以使用另外6个关键词来强化想要生成的服装风格：“casual_clothes”, “japanese_clothes”, “show_clothes”, “loli_clothes”, “leather_clothes” and “china_dress” (对于 V0.2版本)。未来关键词可能会发生变动。可以通过范例图片查看每件衣服的生成咒语。实例内可能有些图片的关键词和版本号可能与V0.2不匹配，不过作为起步点应当是足够了:)希望你喜欢。

[AI Erio] Lora model for hiseki erio/Hiseki Erio (绯赤艾莉欧）

The basic model works fine for me. Use prompts at 0.5~0.8.

Erio_v0.2.safetensors

Erio_v0.2

Files with SHA256: bca54da0da981b9c880e30a9c71210417de7eb9e46296ce730345722953746ae