!!! Before installing custom nodes from manager you need to install this one from Git url
https://github.com/BigStationW/ComfyUI-ppm - negative prompts for Flux1, ZIT and Anima
For example if you don't want to see glasses and tie on Anima, Zit and Flux1 generations add to fixed prompt window string (glasses, tie:-1)
If manager fails to find LLM and lora nodes, install them from git url - https://github.com/willmiao/ComfyUI-Lora-Manager
https://github.com/KingManiya/ComfyUI-LLM-text-processor (Automatic setup currently supports Windows x64 CUDA 13 only. Other platforms require manual setup of llama.cpp release binaries.)
Since v.8.3 there is a separate version of workflow that can be used with Sage Attention enabled in Comfyui launch arguments (it can be easily installed from https://github.com/wildminder/AI-windows-whl?tab=readme-ov-file with no need to install compiler)
This workflow support
Z-Image, Ernie image, MS Lens, Wan, Flux, Flux 2, Flux 2 Klein, Anima, Qwen Image, Qwen Image Edit, SDXL/Pony, Chroma and Lumina-Image 2.0
txt2img, img2img, Inpaint functionality.
Face swap (not that perfect atm)
SeedVR2 upscaler
LLM to describe images and enhance prompts
ControlNet only for Edit models in img2img group (just describe your image in prompt and add some instruction like "use input image as normal map reference")
Outpainting in Edit/Kontext mode
safetensor, gguf and
svdq (Nunchaku)checkpointsPowerful multi Lora loader (Lora manager)
Face After Detailer
Multi or single images load for Edit/Kontext
LLM Models:
VRAM 8 Gb:
VRAM 12+ Gb:
or
Put models and folder with system prompt here:
📂 ComfyUI/
├── 📂 models/
│ └── 📂 LLM/
│ ├── Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf
│ ├── mmproj-Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-f16.gguf
│ └── 📂 prompts/
│ ├── LLM-system.txt
│ ├── LLM-system-SDXL.txtStarting from v8.0, it is possible to enhance T2I prompts on the fly using switches in the T2I group. You still need to manually copy and paste prompts from the LLM group to the I2I group, but you can directly describe Image 1 loaded in the I2I block.
Follow the short block notes for more information.
User prompts in both groups are now divided into two parts. The first part (Fixed) is never passed to the LLM block for processing and will always remain unchanged at the beginning of the prompt. The second part (User Prompt) can be passed to the LLM (if enabled) and improved.
Description
Only pose switch for second reference image (unstable) (6.82)
Mask for image 2 (use as reference only part of image 2) (Fixed in 6.82)
simple style selector (6.7)
navigation hotkeys - try Alt+a and Alt+w (6.7)
Face swap (6.5)
More notes with common prompts (6.6)
LLM model auto download (6.6)
LLM image disable switch (6.6)
FAQ
Comments (15)
Great workflow, thank you. The only problem is that the TS_Qwen_3VL_V2 node is supposedly missing, even though the node pack has been installed (successfully). Since I don't need the function, at least I don't think I do, I'm just ignoring the error message.
Same issue here. Only thing missing.
hmm.. in my last tests TS_Qwen node had no any problems with installing via manager. Going to make more experiments to clarify situation
Update: problem seems to be in latest transformers version. So you need to reinstall it
.\python_embeded\python.exe -m pip uninstall transformers
.\python_embeded\python.exe -m pip install transformers=="4.57.1"
@TikFesku That worked! Thank you!
The LLM part does not function as a pure prompt enhancer without image input, correct?
@CupRunethOverWithLSD It works, but you need to bypass image input node. Because of that there is a group for this single node for easy disabling using icon in top-right corner of group
@TikFesku Excellent! Thank you!
I tried using your workflow today.
After a few restarts, I was able to install almost all the missing nodes with the manager. It's only missing one node: comfyui_hybrid-scaled_fp8-loader. I tried installing this node manually, but the link to the GitHub page results in a 404 error.
I'm using a clean and up-to-date portable version of ComfyUI and your workflow version 6.81.
Can you please help me?
It's confusing. Sometimes the following nodes are shown as missing, sometimes not.
RTXVideoSuperResolution
LTXVSpatioTemporalTiledVAEDecode
LTXVSetVideoLatentNoiseMasks
Can you please point me where in worflow there are LTXV nodes? It must be error - it is image only worflow and supposed to be free from video nodes..
And on my side folder ComfyUI_Hybrid-Scaled_fp8-Loader folder magically appears in custom nodes folder after first run. Just in case i uploaded it here https://drive.google.com/file/d/1aAUhL6Be9zjBEN1RR95eWCLbWPfYRaUS/view?usp=drive_link
You're right.
LTXV isn't mentioned. I think that was an old workflow that was somewhere in my cache.
I cannot download the file. Do you have another download link?
Thank you for this workflow.
I was not able to generate i2i without seedvr2 activate and i find the issue. In your version 6.81 in the node groupe "Shark Sampler i2i", the node "Resolution (CRT)" has the value 1152 in megapixel but the max allow is 32.
Switch the value to 1 fix my generation issue.
Seems that we have different Resolution (CRT) nodes.. On my side longer side have value in pixels - not megapixels. I tried to update this node, gut already have latest version..
Ok now in version 6.83, i have an issue with Upscale Seed VR2 when i used i2i FLux Klein 9b.
The issue is "AttributeError: 'str' object has no attribute 'shape'"
It provide from the node "Src image switch".
Something is wrong with the input Facedetailler and Text2i. If i remove this input, the workflow works well like in the older versions.
@Allerias Hmm.. seems that I need to test it on clean install of Comfyui again.. Thanks for info