Krea2 text to image
Image
krea
3
137
z
zeto
Nodes & Models
UNETLoader
krea2_turbo_fp8_scaled.safetensors
Ver Private
Comm Use
CLIPLoader
qwen3vl_4b_fp8_scaled.safetensors
Ver Private
Comm Use
VAELoader
qwen_image_vae.safetensors
Ver Private
Comm Use
EmptyLatentImage
MarkdownNote
## Krea 2 Turbo 量産用ミニマル構成 **触るのは2箇所だけ** 1. Prompt (English) にプロンプト 2. Resolution / Batch の batch_size で枚数 (既定4) **固定でOKな設定** - steps 8 / CFG 1.0 / euler / simple (Turbo蒸留仕様) - negativeはZeroOut固定 (Turboはネガティブ非対応) **コスト節約** - 1024x1024 (1MP) が最効率。仕上げだけ1536以上に - batch生成の方が1枚ずつよりGPU時間効率が良い **必要モデル (Comfy-Org/Krea-2)** - diffusion_models/krea2_turbo_fp8_scaled.safetensors - text_encoders/qwen3vl_4b_fp8_scaled.safetensors - vae/qwen_image_vae.safetensors
CLIPTextEncode
ConditioningZeroOut
KSampler
VAEDecode
SaveImage
Krea text 2 image workflow
Read more
N
