1.兼容性存在问题，原衣物在部分checkpoint下会出现混乱，如XYZ图所示。2.您可能需要更多的提示词来描述人物所在的场景与环境，（例如：day, blue_sky, cloud, ocean, beach, dappled sunlight）,否则背景可能会出错或变的单调。<h3 id="heading-83">3.人物的衣物可以自定义，在使用自定义衣物时，请降低LORA权重（比如0.65），请保留提示词 katya, aqua_eyes, 1girl, 同时将原衣物类提示词添加进负面提示词，这么做可以改善画风和衣物的过拟合。</h3>4.如果人物的脖子被原衣物提示词过拟合，可以使用collarbone，其他部位也是同理，用更精确的提示词来描绘出现问题的地方，例如bare shoulders, bare arms, bare legs, barefoot5.人物衣物和手部可能会出现错误，特别是在一些特定的动作下，您可以选择移除提示词single_elbow_glove。如果想得到较为精准的还原，您可能需要多次生成，或使用photoshop修复错误6.您可以灵活调整步数、权重，CFG, 默认设置 step:35-40，权重0.9-1 , CFG 7。 启用Hires.fix : R-ESRGAN 4x+Anime6B 放大倍数1.5-2 步数30-35 重绘幅度0.35-0.47.新版本的stable diffusion 增加了LORA提示词与说明的保存功能，可以将提示词保存并快速调用，推荐使用<h3 id="heading-84">8.正面提示词： katya, revealing_skirt, single_elbow_glove, asymmetrical_legwear, choker, hair_ornament, aqua_eyes, 1girl</h3> 负面提示词(请根据实际情况决定是否添加) ：holding_weapon, gun, white_background, 3D rendering ————————————————————————————————————————————1. There is a compatibility problem. The original clothing will be confused under some checkpoints, as shown in the XYZ picture.2. You may need more prompt words to describe the scene and environment where the characters are located (for example: day, blue_sky, cloud, ocean, beach, dappled sunlight), otherwise the background may be wrong or monotonous.<h3 id="heading-85">3. The character's clothes can be customized. When using customized clothes, please reduce the LORA weight (for example, 0.65), keep the prompt words katya, aqua_eyes, 1girl, and add the original clothing prompt words to the negative prompt words, do this It can improve the over-fitting of painting style and clothing.</h3>4. If the character's neck is over-fitted by the original clothing prompt words, you can use collarbone. The same goes for other parts. Use more precise prompt words to describe the problem areas, such as bare shoulders, bare arms, bare legs, barefoot.5. There may be errors in the character's clothing and hands, especially in some specific actions. You can choose to remove the prompt word single_elbow_glove. If you want a more accurate restoration, you may need to generate it multiple times or use photoshop to fix errors.6. You can flexibly adjust the number of steps, weight, and CFG. The default settings are step: 35-40, weight 0.9-1, and CFG 7. Enable Hires.fix: R-ESRGAN 4x+Anime6B Magnification 1.5-2 Steps 30-35 Redraw amplitude 0.35-0.47. The new version of stable diffusion adds the function of saving LORA prompt words and instructions. The prompt words can be saved and quickly recalled. It is recommended to use<h3 id="heading-86">8. Positive prompt words: katya, revealing_skirt, single_elbow_glove, asymmetrical_legwear, choker, hair_ornament, aqua_eyes, 1girl</h3> Negative prompt words (please decide whether to add them according to the actual situation): holding_weapon, gun, white_background, 3D rendering

尘白禁区-凯茜娅 | katya |スノウブレイク：禁域降臨 | SNOWBREAK: CONTAINMENT ZONE | 塵白禁域|

V1.0:1.兼容性存在问题，原衣物在部分checkpoint下会出现混乱，如XYZ图所示。2.您可能需要更多的提示词来描述人物所在的场景与环境，（例如：day, blue_sky, cloud, ocean, beach, dappled sunlight）,否则背景可能会出错或变的单调。3.人物的衣物可以自定义，在使用自定义衣物时，请降低LORA权重（比如0.65），请保留提示词 katya, aqua_eyes, 1girl, 同时将原衣物类提示词添加进负面提示词，这么做可以改善画风和衣物的过拟合。4.如果人物的脖子被原衣物提示词过拟合，可以使用collarbone，其他部位也是同理，用更精确的提示词来描绘出现问题的地方，例如bare shoulders, bare arms, bare legs, barefoot5.人物衣物和手部可能会出现错误，特别是在一些特定的动作下，您可以选择移除提示词single_elbow_glove。如果想得到较为精准的还原，您可能需要多次生成，或使用photoshop修复错误6.您可以灵活调整步数、权重，CFG, 默认设置 step:35-40，权重0.9-1 , CFG 7。 启用Hires.fix : R-ESRGAN 4x+Anime6B 放大倍数1.5-2 步数30-35 重绘幅度0.35-0.47.新版本的stable diffusion 增加了LORA提示词与说明的保存功能，可以将提示词保存并快速调用，推荐使用8.正面提示词： katya, revealing_skirt, single_elbow_glove, asymmetrical_legwear, choker, hair_ornament, aqua_eyes, 1girl 负面提示词(请根据实际情况决定是否添加) ：holding_weapon, gun, white_background, 3D rendering ————————————————————————————————————————————V1.0:1. There is a compatibility problem. The original clothing will be confused under some checkpoints, as shown in the XYZ picture.2. You may need more prompt words to describe the scene and environment where the characters are located (for example: day, blue_sky, cloud, ocean, beach, dappled sunlight), otherwise the background may be wrong or monotonous.3. The character's clothes can be customized. When using customized clothes, please reduce the LORA weight (for example, 0.65), keep the prompt words katya, aqua_eyes, 1girl, and add the original clothing prompt words to the negative prompt words, do this It can improve the over-fitting of painting style and clothing.4. If the character's neck is over-fitted by the original clothing prompt words, you can use collarbone. The same goes for other parts. Use more precise prompt words to describe the problem areas, such as bare shoulders, bare arms, bare legs, barefoot.5. There may be errors in the character's clothing and hands, especially in some specific actions. You can choose to remove the prompt word single_elbow_glove. If you want a more accurate restoration, you may need to generate it multiple times or use photoshop to fix errors.6. You can flexibly adjust the number of steps, weight, and CFG. The default settings are step: 35-40, weight 0.9-1, and CFG 7. Enable Hires.fix: R-ESRGAN 4x+Anime6B Magnification 1.5-2 Steps 30-35 Redraw amplitude 0.35-0.47. The new version of stable diffusion adds the function of saving LORA prompt words and instructions. The prompt words can be saved and quickly recalled. It is recommended to use8. Positive prompt words: katya, revealing_skirt, single_elbow_glove, asymmetrical_legwear, choker, hair_ornament, aqua_eyes, 1girl Negative prompt words (please decide whether to add them according to the actual situation): holding_weapon, gun, white_background, 3D rendering

katya-v1.safetensors

v1.0

Positive prompt words: katya, hair ornament, aqua eyes, revealing skirt, single elbow glove, asymmetrical legwear, chokerNegative prompt words (if needed): holding weapon, gun, white background, 3D renderingWhen using customized clothes, please reduce the LORA weight (for example, 0.65), keep the prompt words katya, aqua_eyes, 1girl, and add the original clothing prompt words to the negative prompt words

Files with SHA256: 4ec074d63ee6ca1d9b4d868599876863ed51d5461f6e818aa3d2bffbf1b99ede