Z-image-Turbo ControNet Pose-Avanced
Pose-Guided Image Generation with Z-Image-Turbo + ControlNet Union 2.1 Generate realistic images that follow the exact pose of your reference photo. This workflow extracts pose data using DWPose, then generates a new image matching that pose through Z-Image-Turbo and ControlNet.
comfyui
controlnet
img2image
pose
zimage
zimage-turbo
1
199
Nodes & Models
UNETLoader
z_image_turbo_bf16.safetensors
ModelPatchLoader
Z-Image-Turbo-Fun-Controlnet-Union-2.1.safetensors
ModelSamplingAuraFlow
CLIPLoader
qwen_3_4b.safetensors
LoadImage
CLIPTextEncode
ImageScaleToMaxDimension
ConditioningZeroOut
GetImageSize
EmptyLatentImage
PreviewImage
QwenImageDiffsynthControlnet
KSampler
VAEDecode
SaveImage
Image Comparer (rgthree)
AIO_Preprocessor
AIO_Preprocessor
Upload a reference image → DWPose extracts the body pose → Z-Image-Turbo generates a new image guided by that pose via ControlNet Union 2.1. Qwen 3 4B handles text encoding for your prompt, and the turbo model delivers results in 10 steps at CFG 1.0 — fast enough for real-time iteration.
The pipeline includes ImageScaleToMaxDimension for consistent input sizing, ConditioningZeroOut for clean negative conditioning, and ModelSamplingAuraFlow for optimized sampling. Image Comparer node lets you review reference vs. output side by side, and the upscale node brings your final image to production-ready quality.
Drop in your reference, write your prompt, hit run.
Read more
N