PVC Figure Style for Asanagi's work (for Illustrious-XL) | 朝凪手办风

2024/12/25 update:use highfix or ADetailer if character's face get blurry or twisted.Recommand Weight: 1.1 ~ 1.5Trigger Word:asanagi-figure,Some Illustrious-XL based models have stong tendency to draw in anime style rather than in realistic style (for example, Obsession), for those models, increase this lora's weight accordingly, and you may also need to put trigger words like:(((asanagi-figure, photorealistic, photo (medium), realistic))), Fixed the problem that v1 will get blurry on recent illustrious model (for example, Noob 1.1)Add a new feature "cum on figure": use tag like:((cum on figure)), can generate realistic semen flooding on figure, but need cum related tags and will need to do some gachaOther Useful tags for cum on figure : cum,cum overflow, bukkake, cum on body, facial, etc.建议使用高清修复或者ADetailer来修复人物脸部模糊或者扭曲的问题推荐权重：1.1~1.5修复了v1无法兼容最近较新的illustrious模型的问题，此外增加了射手办上的新功能，能生成某些变态的返图一样的效果。。。不过需添加精液相关tag并多roll几次。Trained specifically for figure works originally Illustrated by Asanagi, trained on Illustrious-XL 0.1Also works on several Illustrious-XL based model (for example, NoobAiXL, Lucereon), NoobAiXL seems to have best result so far in my test.Recommand Weight: 1.0 ~ 1.2Trigger Word:asanagi-figure,Some Illustrious-XL based models have stong tendency to draw in anime style rather than in realistic style (for example, Obsession), for those models, increase this lora's weight accordingly, and you may also need to put trigger words like:(((asanagi-figure, photorealistic, photo (medium), realistic))), Most data was collected on Twitter.Old version for Animagine V3 & V3.1:PVC Figure Illustrated by Asanagi (for AnimagineXLV3 and AnimagineXLV3.1) | 朝凪手办风 - v2.0 for ani3.1 | Stable Diffusion XL LyCORIS | Civitai（Because the usage is different so I open this new post）专门为朝凪的手办化作品训练的画风，底模为Illustrious-XL 0.1, 其他基于Illustrious-XL 0.1的微调模型理伦上也能生效(比如NoobAiXL, Lucereon)，NoobAiXL在我的测试里表现是最好的。推荐权重：1.0~1.2触发词：asanagi-figure,部分基于Illustrious-XL 0.1的微调模型有很强的动漫画风倾向（比如 Obsession），在这类底模上需要提高这个lora的权重，同时你可能需要这么写触发词：(((asanagi-figure, photorealistic, photo (medium), realistic))), 绝大部分数据收集自推特的买家返图。针对Animagine V3 & V3.1训练的旧版模型：PVC Figure Illustrated by Asanagi (for AnimagineXLV3 and AnimagineXLV3.1) | 朝凪手办风 - v2.0 for ani3.1 | Stable Diffusion XL LyCORIS | Civitai（主要是因为用法不同所以我重新开了这个新的post）