2d 国漫/Chinese cartoon

感谢使用，

推荐权重：                   0.6-0.9      我用的0.8

我目前测试的底模：     dalcefoPainting_3rd，，其它的自测哈，

                                    底模链接 https://ko-fi.com/s/9dd78df21d ，2d底模都可试试，

embeddings模型：     EasyNegative，ng_deepnegative_v1_75t，badhandv4

VAE:                             vae-ft-mse-840000，animevae， 如果底模颜色过重可换vae

Sampler:                     DPM++ SDE Karras,   DPM++ 2M Karras，Euler a

CFG scale:                   7

步数： 20-40，

出全身要用大尺寸图，否则脸崩，我用的是512x768，高清修复放大2倍1024x1536,或更大，不要同时用脸部修复，

建议只开高清修复，

触发词是没有的。。不过要出元素，由于lora中的元素包含的元素有限，所以很看底模发挥，以下这些词仅供参考，

long hair,boots,lotus shoes,chinese clothes,hanfu,Ink background, ink painting, mountain, plum blossom, pavilion,hair ornament,long sleeves, wide sleeves, jewelry,

长发，靴子，荷花鞋，中国服装，汉服，水墨背景，水墨画，山，梅花，亭子，发饰，长袖，宽袖，首饰

——————————————————————————————————————————————————————————

Thank you for using

Recommended weight: 0.6-0.9, I'm currently using 0.8.

The base model I'm currently testing is dalcefoPainting_3rd, and I've also tested other models.

Base model link: https://ko-fi.com/s/9dd78df21d. Lora is highly versatile and can be used with 2D base models.

Embeddings models: EasyNegative, ng_deepnegative_v1_75t, badhandv4.

VAE: vae-ft-mse-840000, animevae. If the base model has heavy colors, you can switch to VAE.

Sampler: DPM++ SDE Karras, DPM++ 2M Karras, Euler a.

CFG scale: 7.

Number of steps: 20-40.

For full-body images, use large-sized images, otherwise the face may break. I use 512x768 images, and upscale them to 1024x1536 or larger using high-definition restoration, but do not use face restoration at the same time.

It is recommended to only use high-definition restoration.

There are no trigger words, but to generate elements, the options are limited in Lora, so it depends on the performance of the base model. The following words are for reference only: long hair, boots, lotus shoes, Chinese clothes, Hanfu, ink background, ink painting, mountain, plum blossom, pavilion, hair ornament, long sleeves, wide sleeves, jewelry.