3D模型丨可爱化SDXL版版本v2.0 (ID: 177483)

The cute sdxl version is updated, mainly for the texture of the output model, the texture of hair, clothing and other materials; Greatly reduce the model memory; Among them, the higher the weight of the model, the higher the animal is recommended; (High resolution fixes are best turned on) the probability of drawing will be more stable. More gameplay you can explore on your own.

The 3D model based on SDXL training is similar in style to the Pixar style, but the characters are more inclined to the blind box style. In addition to characters, it can also generate some elements such as animals and items. The probability of drawing is more stable. More gameplay you can explore on your own.

Welcome to post and share

It is recommended to use the Stable Diffusion SDXL official version for higher compatibility.

Size :768*1024 or 1024*1024, sampling method recommended DPM++ 2M Karras

Settings: Sampling: 35 steps, weight :0.8-1.2, DPM++ 2M Karras, CLIP: 2

Trigger words: MG_ip, pixar

V2.0:

Hair texture: ++++

Clothing, metal, and other textures: ++

Truth degree: +

Saturation: +

Occupied memory: --

Mainly for the output model texture improvement, hair, clothing and other materials texture; Greatly reduce the model memory; Among them, the higher the weight of the model, the higher the animal is recommended; (High resolution fix is best turned on)

描述:

V2.0:

毛发质感：++++

衣服、金属、等质感：++

真实度：+

饱和度：+

占用内存：- - -

主要对于输出模型质感提升，毛发、衣服等材质的质感；大幅缩减模型内存；其中模型权重越高人物约q版化，动物推荐高一点；（高分辨率修复最好打开）

训练词语: MG_ip