Instagram: https://www.instagram.com/synth.studio.models/
Buy me a☕ https://ko-fi.com/lonecatone
This represents hundreds of hours of work. If you enjoy it, please 👍like, 💬 comment , and feel free to ⚡tip 😉
Note: If you have ANY issues with nodes not downloading, read the notes or reach out. There's nothing that special about any of them that aren't core modules.
This is an advanced workflow. I encourage you to take the time to learn it. If you feel overwhelmed, you can start here:
https://civarchive.com/models/2348914/zimage-base-and-turbo-essentials-workflow
A complete user and troubleshooting guide inclusing a list of common error messages can be found here https://civarchive.com/articles/24678/how-to-use-my-workflows-a-comprehensive-breakdown
也有中文说明
V10.0
Upgraded Samplers to Clownshark Samplers
Added a Denoise Slider
Added the ability to color correct to the orgiginal image without having to load it again
minor additions
V9.0
Because it is not in my nature to leave things alone....
Gutted the stock face detailer nodes and replaced them with a SEGS suite for infinately better results both SFW &NSFW
Updated links to teh latest detailers for all
Controlnet Update.
Added local bypass nodes so ou don't have to zoom and scroll
Added a quality modifier prompt area that follows after Qwen so that they stay preserved.
Major update to notes, especially for detailers
Ver 6.0
Overhauled the detailers
Body detailers are better calibrated
Face. eyes, and hands are now run by Z-Image for better consistency
Please give me feedback with any issues on this, especailly RAM related.
added a LoRA area for character and detail LoRAs
Changed the save area:
You can choose to embed the workflow or not
You can add a simple date to the metadata folder for better organization
Changed the prompt area:
added a sentence remover that helps eliminate unwated words (CivitAI flags 🚩)
Added an inline crop tool in the post processing Suite
Minor correction to the gguf area to deal with the quantized DIT models crashing.
Added better notes.
Ver 5.0
Added Face Swap
Added a sentence remover to prevent CivitAI flags
Improved detailers
Fixedbroken metadata saver from ComfyUi
Ver 4.0
I did a massive overhaul to catch it up to the new nodes and ComfyUi updates that broke a lot of stuff.
Upgraded The Controlnet and prompt assist area
Added a sampler swap to make it easier to go from Base to Turbo wityhout having to reset everything.
Major overhaul of the detailer area (see release notes)
Major overhaul of the post production Suite to focus on realism/ camera
All in 1 workflow:
Controlnet
Lora Loader
Advanced KSampler
Seed variance enhancer
Ultimate Upscaler for prescaling and hi res fix
Detailing Suite for face, hands, eyes, and expressions
⛔ NSFW anatomy correctors (Male and Female)
Anime detailer
Seed VR2 Upscaler to get 4k on low VRAM
Massive Post Production Suite
也有中文说明
All images were created using the standard FP16 version with no loras (except for a character one). For even better quality, use this Lora with it: https://civarchive.com/models/2206377/zit-mystic-xxx
Check out my other models:
Original Version (SFW): https://civarchive.com/models/2184844/zit-workflow-with-prompt-assist-controlnet-lora-loaders-inpainting-upscaler-detailers-expression-control-and-post-processing-i2i-t2i
For ultimate Anime detail and control: https://civarchive.com/models/2220766/zimage-ultra-anime-detail-workflow-get-the-most-out-of-your-generations
For Realism and variety over the base model: https://civarchive.com/models/2231181/z-image-ultra-real-workflow

Description
Added Face swap
Made changes to fix broken ComfyUI metadata
Improved Detailers
Added a Flagged sentance remover for CivitAi prompts
FAQ
Comments (5)
Heya Lonecats,
I'm running this flow on a clean Comfyu Portable Installation. I have all the models, loras, nodes, et al.
Couple of errors I am seeing. I'm working them but thought I've give the feedback and fixes for any other users who may be seeing similar issues.
Switching these nodes off lets things flow as expected - fantastic output and great job, and thanks for all this hard work!
Learning this AI stuff using your workflows is a fun challenge! :)
Florence2Run
AssertionError: only support square feature maps for now
- I thought that the resize node just below this section would handle that but I'm still following the spaghetti. :)
QwenVL-Mod Prompt Enhancer
ValueError: Couldn't instantiate the backend tokenizer from one of: (1) a tokenizers library serialization file, (2) a slow tokenizer instance to convert or (3) an equivalent slow tokenizer class to instantiate and convert. You need to have sentencepiece or tiktoken installed to convert a slow tokenizer to a fast one.
- I have sentencepiece and tiktoken installed and updated - Still working this one.
ReActor
2026-04-07 08:00:54.3399077 [E:onnxruntime:Default, provider_bridge_ort.cc:2359 onnxruntime::TryGetProviderInfo_CUDA] E:\_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1952 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : Error loading "C:\AI_Apps\Comfy_Z-Image\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll" which depends on "cublasLt64_12.dll" which is missing. (Error 126: "The specified module could not be found.")
- Fixed by installing Cuda 12.1 runtime into the python environment and copying cudart64_12.dll from "..\python_embeded\Lib\site-packages\nvidia\cuda_runtime\bin" to "..\python_embeded\"
- Going to install the runtime system wide as that will be MUCH easier for my other Comfy installations that may require it - I don't think moving the DLL will be needed when installed system wide.
Anyhoo! Great work, once again!
You shouel be using Florence Prompt gen base or large 2.0. That's the wrong model error
Qwen VL shoudl no have issues if you are using 2b or 4b instruct. Again I think you may have teh wrong model selected
Florence is broken. Hopefully they update.
Mother fucker. I literally just pulled Qwen out of my workflows
@lonecatone23 "Florence is broken."
Doesn't that always figure. :P
@maxoverride871 yeah, i's always something. Easy fix though. Just insert a resize node and make it 512x512 for analysis













