✨ FLUX — UltimateSD Upscale — Simple Workflow
A clean, all-in-one FLUX UltimateSD Upscale workflow built entirely with the UmeAiRT Toolkit for ComfyUI.
Only 6 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.
⚠️ IMPORTANT — Nodes 2.0 Required
This workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.
How to activate Nodes 2.0:
Open ComfyUI
Go to Settings (⚙️ icon, bottom-left)
Find "Use Nodes V2 (Vue)" and toggle it ON
Refresh the page
Load the workflow
If you prefer the classic interface, check out my Legacy version of this workflow instead (link).
📦 Custom Node Required
Only one custom node to install:
Install via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.
The Toolkit packages everything internally — upscaler, face detailer, metadata saver. No other custom nodes needed.
📂 Files you need (in manual version)
for "base" version :
Model : flux1-dev-fp8.safetensors or flux1-dev-fp16.safetensors
in ComfyUI\models\diffusion_models
CLIP : clip_l.safetensors
in ComfyUI\models\clip
for GGUF version :
Model : Q8, Q6, Q5, Q4, Q3
in ComfyUI\models\unet
CLIP : Q8, Q6, Q5, Q4, Q3
in ComfyUI\models\clip
Common Files :
Text encoder : t5xxl_fp8_e4m3fn.safetensors or ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors
VAE : ae.safetensors
in ComfyUI\models\vae
Description
Tested on ComfyUI v0.3.62.
A few modifications to standardize my workflows.
FAQ
Comments (3)
3.0 Works perfectly. but 3.1 just right away give me a lot of errors
- Return type mismatch between linked nodes: conditioning, received_type(CLIP) mismatch input_type(CONDITIONING) CLIPTextEncode: - Return type mismatch between linked nodes: clip, received_type(STRING) mismatch input_type(CLIP) GetImageSizeAndCount: - Return type mismatch between linked nodes: image, received_type(CONDITIONING) mismatch input_type(IMAGE) DownloadAndLoadFlorence2Model:
........
and the list just incrieses for what i see the Florence model have all the input mixmatch but after comparing it to the 3.0 they seem similar sooo no clue, and they will still get trigger if i disable automatic promt
Same thing for me. i've got the same list of issues. 3.0 work fine
I've got better upscale result with the workflow Qwen img to img and this lora https://civitai.com/models/2046703?modelVersionId=2368038