Welcome to KREA 2 Diffusion Workflow by Blackfish95
This page has been succeeded by my new v4.0 WF.
The difference is minimal, the main purpose was to provide new Krea2 users a clear set of guidelines for how to get up and running with Krea 2 on ComfyUI.
Try my Model!
Required Node Packs: [Use Comfy Manager search]
ComfyUI-Lora-Manager
RES4LYF (For better Samplers/Schedulers)
RGThree Nodes (Recommended but not needed for the workflow)
Suggested Sampler/Schedulers (Try these or check Model page for what to use)
euler/simple (baseline)
euler_a/beta
euler_a/bong_tangent
er_sde/bong_tangent
er_sde/beta
er_sde/sigmoid_offset
DPM++ 2M SDE / Simple
Model Links (for Local Users)
For BF16 and NVFP4 variants, see
Krea 2 Original Model
text_encoders
Model Default - qwen3vl_4b_fp8_scaled.safetensors
vae
Model Default - qwen_image_vae.safetensors
Enjoy and please share your creations on this page. I love to see what you create!
Description
My workflow has been updated with some new nodes to help with LoRA management, and Civitai compatible image saves.
Install Pixaroma and ComfyUI-Lora-Manager first to make it easier to load and run this workflow.
Required Node Packs: [Use Comfy Manager search]
ComfyUI-Lora-Manager
Pixaroma
Recommended but not needed for the workflow: RGThree Nodes
Sampler / Scheduler suggestions
euler / simple
euler_ancestral / bong_tangent (My current preferred)
er_sde/beta
DPM++ 2M SDE / Simple
I leaned heavily on some of the features of the Pixaroma node pack. Including node colours and the timer node with a 'job complete' alert. You can just delete that timer node without breaking anything, but you can also adjust the sound volume or select a different sound.