Descriptions:
The Nerdy Rodent channel presents a detailed tutorial on Union Control Net 2.1 for Z Image Turbo in ComfyUI, covering an updated control node and newly added inpainting capabilities. The video continues from prior Z Image Turbo tutorials and focuses on three new workflow groups: the updated Union 2 control net, Union inpaint, and prompted Union inpaint.
A key change is the replacement of the older Quen node with the new Z Image Fun Control Net node from Comfy, which adds dedicated inpaint image and mask inputs. The tutorial covers practical strength settings — 0.7 for pose control guidance, and 0.8–0.9 when using empty latent for inpainting — and explains the trade-off between the two inpainting methods: empty latent with high strength preserves non-masked colors but degrades the inpainted region, while image encode with lower strength and set latent noise mask achieves strong results in both areas simultaneously. Differential diffusion integration is also explained as an optional enhancement.
For users who prefer automated masking, the tutorial demonstrates SAM2 segmentation from the ComfyUI RMBG pack as a prompt-based alternative to manual brush masking. A free downloadable workflow is available via the creator’s Patreon. The video is aimed at existing ComfyUI users familiar with Z Image Turbo who want more precise inpainting and pose control in their image generation pipelines.
📺 Source: Nerdy Rodent · Published December 20, 2025
🏷️ Format: Tutorial Demo
