This beginner friendly Z-Image Turbo ALL-in-One Controlnet Union workflow can generate a new output image using a reference image for pose and style guidance (Canny/Depth), while applying detailed text prompts to control the final look of desired output. This is easier to follow for any early adopter of ComfyUI as it does not use sub-graphs or too complex processes. Make sure that you have latest enough versions of both ComfyUI and ComfyUI manager to manage and install any missing dependencies (missing nodes, patches etc.) to use this properly. This workflow by default uses Fp8 version of the Z-Image Turbo ALL-in-One Model which has Clip & VAE backed in, this will work well on base level Apple Silicon Mac M processors (like M3, M4, M5 etc.) with only 16GB unified memory or on a PC with low end Nvidia and AMD GPUs/eGPUs.. even on 8GB VRAM. But if you want to use bigger bf16 version of the Z-Image Turbo ALL-in-One Model on a more powerful system I also provided links in the readme file. Negative prompts will only work on this workflow if you use CFG of 1.2 and Steps 12 or higher (I have found CFG 1.2 & 12 Steps combo producing better results) to filter out what you don't want to see on your outputs (use the defaults of this workflow for optimized results, only change if you know exactly what you want). In the AIO Aux Preprocessor node, the following preprocessors have been confirmed to work well and are highly recommended for optimal results:
# CannyEdgePreprocessor - For generating output with a Canny Reference Image
# DepthAnythingPreprocessor - For generating output with a Depth Reference Image
# OpenposePreprocessor - For generating output with a Pose Refenence Image
# DWPreprocessor - For more accurate output than the standard OpenPose for body & hand tracking
How to use this -
#1. Just select your desired Z-Image Turbo model first and then
#2. Load your input reference image file (.Png or .Jpg or anything supported )
#3. then input your prompts (positives and negatives).
#4. Now you may select an appropriate AIO Aux Preprocessor ( Canny, Depth, Openpose.. or just use the default - DWPreprocessor )
#5. select how many images you want ( Change the number besides the "Run" button )
#6. select image sampling methods, cfg, steps etc. settings ( Default values - CFG 1.2 & 12 Steps gives most optimized outputs )
#7. finally press the run button to generate. That's it.
*** If you are using a ComfyUI managed via Stability Matrix then make sure you download and put the Z-Image Turbo Controlnet Union model patch inside models subfolder of ComfyUI package subfolder itself, not the Stability Matrix's common models subfolder.. so it will go to <Your Stability Matrix's Data folder>/Packages/ComfyUI/models/model_patches not into <Your Stability Matrix's Data folder>/Models/model_patches (the shared common model's Subfolder where you commonly store models so they can be shared between ComfyUI, WebUI Forge etc ..multiple different tools you may install and manage via Stability Matrix). If you can't do this correctly this workflow will not work.
## Required Models
===================
### Download Links for Z-Image Turbo All-in-One Checkpoints (with baked in Clip & VAE)
https://huggingface.co/SeeSee21/Z-Image-Turbo-AIO/resolve/main/z-image-turbo-fp8-aio.safetensors
https://huggingface.co/SeeSee21/Z-Image-Turbo-AIO/resolve/main/z-image-turbo-bf16-aio.safetensors
### Download Link for Z-Image Turbo Controlnet Union model patch