Four Krea2 workflows that take the model from 'generates nice images' to 'generates the image I specified' — spatial control, detail enhancement, and a prompt generator.

The four workflows

# Workflow Purpose
1 RAW + Turbo high/low noise sampling Two-pass sampling: RAW for structure, Turbo for finish
2 Image edit + detail enhancement Edit an existing image and lift local detail
3 Depth map spatial control Lock composition with a depth map
4 Random prompt generator Produce structured prompts for batch exploration

The part that actually matters

Depth control is the one that changes how the model behaves. Without it, every regeneration is a
new composition. With a depth map, the layout is pinned and the model spends its capacity on
surface instead of on deciding where things go.

The high/low noise split is the second useful piece: low-noise pass for composition, high-noise
pass for texture, which avoids the plastic sheen you get when a single pass does both.

Running Hub links

Files

Workflow JSONs and covers: https://huggingface.co/AI-KSK/ksk-video-275-comfyui-workflows

Video walkthrough (Chinese): https://youtu.be/pfGawt0ygHA

Bilibili: https://www.bilibili.com/video/BV1nxM56cEta/

Notes

  • ComfyUI exports that reference models present in the cloud instance; not self-contained packs.
  • Tutorial material, not a product.
  • Independent AI-KSK release. Not affiliated with or endorsed by the model authors or RunningHub.