TAG: ganyu \(genshin impact\), 1girl, solo, sblack gloves, bare shoulders, breasts, elbow gloves, sketch, neck bell, blue dress, alternate costume, detached sleeves,
Training model: Animagine XL V3
Inference model: Animagine XL V3
Training resolution: 1024*1024
Recommended output resolution: 1024*1280
Recommended model weight: 0.8~1
新衣服的素材不多，相比旧服装表现稍差一点
Description
FAQ
Comments (9)
可以让其他原神角色穿上这套衣服（误）
very well done lora
do you have your training settings anywhere to view or would be willing to share them?
您可以通过SD-webui的lora界面去查看他，我没有特意隐藏。
thank you for the response!
i notice on one LorA you have "buckets" enabled and another disabled. Whats the reason for this?
thank you again
我在中文互联网分享了我的训练配置，但我不建议你直接借鉴这些参数，而是根据自己的实际情况灵活调整：https://drive.t4wefan.pub/d/windsing/wingsing/example_SDXL_lora.json
@rippinao556 我应该所有的lora训练都开启了bucket，以便适应不同分辨率大小的训练素材。但是也有可能某些lora在bucket上没有做好调整，我注意到最近有些lora会有小概率发生“头部丢失”的情况，我不确定是否与这有关。如果您方便的话，可以将有疑问的模型链接发我，我会重新检查他们
@Chenkin thanks for the swift response. how do you decide how many repeat images for training vs epochs? And if you use wd14 captioning what are your values for "general threshold" and "character threshold"? you have been a great help, thank you!
@rippinao556 假设您有500张申鹤的图片 ，那么repeat为1,epoch是30∽50，是比较合适（注意，如果你要学习的角色比较复杂，请适当拉高步数）。使用wd14打标的话，我一般用默认值——大概是0.35.
@Chenkin thank you for your help! looking forward to your next lora !
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.