Scraed adds Z Image Base support for LanPaint in ComfyUI

By vramkickedin | | 2 min read
A display of the custom LanPaint ComfyUI nodes

Developer scraed released an update to LanPaint on January 30, 2026, adding documentation and workflow support for the Z-image-base model. This update expands the tool's compatibility list, which already includes popular models like Flux, SDXL, SD 3.5, Hunyuan, Wan 2.2, and Qwen Image Edit.

The extension functions as a universal inpainting sampler for ComfyUI, allowing users to perform inpainting tasks without additional model training.

LanPaint's core features & capabilities

  • Model compatibility with Z-image, Hunyuan, Wan 2.2, Qwen Image/Edit, Flux, & more.
  • Training-free operation that functions with existing models immediately.
  • Support for any mask shape, size, or position for inpainting and outpainting.
  • Content generation without relying on blending or partial denoising workarounds.
  • Video inpainting and outpainting capabilities with Wan 2.2 support.
  • Character consistency generation through Prompt First mode.

Z-Image-Base Integration Details

The new Z-image-base support comes with specific usage recommendations from the developer. According to the GitHub documentation:

'Z-image-base can easily diverge with LanPaint. Start with small LanPaint_StepSize and fewer thinking iterations (lower LanPaint_NumSteps) and increase gradually only if stable.'

That said, users can access the workflow template directly through the provided Z_image_base_Inpaint.json file. The recommended setting for Z-image-base inpainting uses the LanPaint K Sampler with 3 steps of thinking.

Check out LanPaint?