Instructions to use wikeeyang/Krea2-Turbo-HD-V1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use wikeeyang/Krea2-Turbo-HD-V1 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("wikeeyang/Krea2-Turbo-HD-V1", dtype=torch.bfloat16, device_map="cuda") prompt = "这是一幅极具东方古典韵味的特写镜头作品,画面采用平视视角,聚焦于一位身着华服的少女侧面半身。 光线柔和地打在人物面部,营造出如水彩般温润的质感。主体少女以侧颜入画,头部微倾,眼神流转间流露温婉与一丝羞涩。 她的发髻高挽,深青色的秀发间点缀着繁茂的金桂与翠绿叶片,几朵金黄的桂花垂落鬓角,与发丝纠缠,发间还隐约可见 精致的金边流苏步摇。她的妆容精致典雅,眉如远山,眼尾微挑,卧蚕饱满,唇瓣虽未完全显露,却似含情脉脉。她身着 一袭青碧色与米白相间的汉服,衣料轻薄,纹理间隐约可见暗纹云锦。右手优雅地持握着一柄绘有工笔花鸟的团扇, 扇面轻掩下半张脸庞,仅露出一双含情目。她手腕处佩戴着一只繁复的古典手钏,绿松石与金珠交相辉映,流苏轻垂。 画面色彩滤镜偏向低饱和度的复古青绿调,背景为留白式的淡雅色调,边缘处有水墨晕染的痕迹,使得主体更加突出。 构图上,人物的面部与团扇形成了巧妙的视觉三角,发间的枝叶与衣袖的线条相互呼应,充满了写意与工笔结合的细腻风格。" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
[update it works]how to load Krea2-Turbo-HD-V1-int4_convrot.safetensors in Comfy?
how to load Krea2-Turbo-HD-V1-int4_convrot.safetensors in Comfy?
I tried newest krea2-turbo-INT8 workflow but failed.
note: in this workflow your Krea2-Turbo-HD-V1-INT8_convrot.safetensors is working.
how to load Krea2-Turbo-HD-V1-int4_convrot.safetensors in Comfy?
I tried newest krea2-turbo-INT8 workflow but failed.
note: in this workflow your Krea2-Turbo-HD-V1-INT8_convrot.safetensors is working.
I think native inference is already in ComfyUI, but I haven't tested it; instead, I have preferred to use these custom nodes, and they work well for me. If something fails for you, try them, or check if they have the same speed as the ones you are currently using. https://github.com/viralvfx/ComfyUI-INT4-Fast
how to load Krea2-Turbo-HD-V1-int4_convrot.safetensors in Comfy?
I tried newest krea2-turbo-INT8 workflow but failed.
note: in this workflow your Krea2-Turbo-HD-V1-INT8_convrot.safetensors is working.
updating your comfyui, I suggest.
yes! I saw this update today, it now works

