CIVITAI / Workflows

Z-Image-ControlNet + SDPOSE + second-order Refiner

This workflow uses the enhanced Z-Image Turbo model together with ControlNet to achieve stable control over lines, depth maps, and pose skeletons, covering nearly all common conditioning types. Z-Image Turbo remains extremely fast while respecting the structural guidance extracted from SD-Pose or other detectors, making the first-pass generation both pose-accurate and responsive to the control signal. Because ControlNet naturally lowers aesthetic richness, the workflow adds a secondary refinement stage: the image is decoded, re-encoded into latent space, and passed through an additional diffusion round. This brings back detail, color depth, and overall visual quality without breaking the control. Latent-space upscaling often produces more pleasing variation compared with direct pixel upscaling, so the workflow uses it as the primary refinement strategy. In practice, results show that skeleton control is stable and predictable, and switching to SD-Pose improves the accuracy of complex or full-body detections. For more stylized or line-art inputs, preprocessing strength directly influences how strictly the model interprets the structure, and adjusting it provides smoother control than relying solely on ControlNet’s strength value. ControlNet’s presence can weaken LoRA effects, so users may choose carefully when applying style models. Depth extraction, line control, and even typography guidance work reliably, though aesthetic quality still benefits from the two-stage refinement. Given that the Z-Image ecosystem is new and evolving, this workflow provides a practical balance: strong structural control, fast generation, and a refiner step that restores the vivid aesthetics Z-Image is known for. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/7GtFnr8ixnM Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/199659