如果您需要定制lora，你可以通过QQ：3692722976添加我If you need a custom LoRA, feel free to DM me on CivitAI. I'll reply as soon as I see your message!----------------------&gt;This LoRA model is a test focused on training a multi-character model for Anima. The characters featured are Sandrone and Columbina from &lt;Genshin Impact&gt;.Tested and confirmed: this LoRA is capable of generating consistent multi-character interactions without character bleed or feature confusion.The training dataset consists entirely of tags. While it does not include natural language captions, testing shows that this LoRA responds well to natural language prompting. You can accurately reproduce the characters' appearances using only their trigger words. To change their outfits, simply add the corresponding clothing descriptions to your prompt.Of course, it works perfectly for generating single characters as well.这次的模型测试的是如何训练anima的多人lora，角色是来自《原神》中的桑多涅（木偶）、哥伦比娅（少女），训练集为纯tag训练集，虽然无自然语言标签，但经过测试，此lora也可以用自然语言触发。此lora仅使用触发词就可以还原人物外貌，如果需要改变角色服饰，加上对应服饰的描述即可,经过测试，此lora可以跑出双人互动的效果，且角色特征不会出现混乱，当然，它也可以跑出单个角色。----------------------&gt;To help you get the best results, I have also uploaded the workflow used for testing this LoRA. This workflow can be used for standard image generation needs with the Anima model.You can find the VAE and CLIP for the Anima model on its official release page.<a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima">https://civarchive.com/models/2458426/anima</a>为了更好的让用户使用这款lora，我同时上传了用于测试这个lora的工作流，此工作流可以用于anima模型的正常跑图需求。 anima模型的vae和clip可以在官方的发布页面获取： <a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima">https://civarchive.com/models/2458426/anima</a> ----------------------&gt;Follow me for more models—thanks! ♪(´▽｀)

Files with SHA256: 28e6c0ef68be467e89056bbe4097cb00e60bc5eb166d2fc201e9744f4e3ea25b