The lora is suitable for Internet operation, children's illustration and other scenarios
The lora style characteristic is relatively strong, applied in the second element, vector illustration is also applicable. This model is relatively stable, but needs to be adapted to the anything_v5 larger model. It is recommended to turn on the ADetailer when the face is broken
Recommended parameter
Training base mode: anything_v5
Trigger word: None
Recommended weight: 1
Recommended checkpoint: anything_v5
Recommended resolution: 512x512, 768x512
Recommended VAE: animevae or automatic (not recommended vae- FT-MSE-84000-EMA-Pruned)
Recommended sampling method: All can be (different samplers, different styles, but the effect is still good)
Hd repair algorithm: Practical HD repair will make the picture distorted
Welcome to return to the picture and attention, your attention is the biggest motivation for my creation! ~
该lora适用于互联网运营、儿童插画等场景使用
该lora风格特征性比较强，应用在二次元，矢量插画也适用,本模型相对比较稳定，但需适配anything_v5大模型,在面部崩坏时建议打开ADetailer
推荐参数
训练底模：anything_v5
触发词：无
推荐权重：1
推荐搭配大模型：anything_v5
推荐分辨率：512x512、768x512
推荐VAE：animevae或者自动（不推荐vae-ft-mse-840000-ema-pruned）
采样方法推荐：均可（不同的采样器，风格各有千秋,但是效果还是都不错的）
高清修复算法：实用高清修复会让画面扭曲
欢迎大家返图和关注，你的关注就是我创作的最大动力！~
描述:
训练词语:
名称: small_scene_illustration.safetensors
大小 (KB): 73843
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
