PixaromaHelp
Help › Nodes › Utility

Krea LoRA Converter

Make a fal.ai Krea 2 LoRA work in ComfyUI.

In ComfyUI: double-click an empty spot on the canvas and type Krea LoRA Converter.

What it does

LoRAs you train on fal.ai for the Krea 2 model use layer names ComfyUI does not recognize, so they will not load. This node renames those layers to the names ComfyUI expects and saves a new copy in your loras folder.

Nothing about the training changes. The weights are copied across exactly, so the result is identical, just loadable. The new file is a normal ComfyUI LoRA: load it with any LoRA loader, in any workflow, alongside the Krea 2 model.

How to use

  • Pick your fal Krea 2 LoRA from the dropdown.
  • Read the line below the button. It confirms it is a Krea 2 LoRA and shows the name it will save.
  • Click Convert. The new file appears in your loras folder and in every LoRA dropdown.
  • Type a different Output name first if you like, or turn on Overwrite to replace an existing file.

Good to know

  • It only reads your file and writes a new one. It never changes or deletes the original.
  • It never downloads anything.
  • If a layer is not recognized it tells you, instead of quietly dropping it.
  • You can also just press Run to convert, but the Convert button is quicker.

The settings on the node

lora_namechoice
The LoRA to convert. Pick a Krea 2 LoRA you trained on fal.ai. The readout on the node confirms whether it is a fal Krea 2 LoRA before you convert.
output_nametext
Filename for the converted copy (saved next to the original). Leave blank to use the original name plus _comfyui.
overwriteon / offdefault off
If a file with the output name already exists, replace it. Off by default so you never overwrite a file by accident.

What comes out

Nothing. This is the end of a chain - it shows or saves what reaches it.

Independent tool. Not affiliated with or endorsed by Krea or fal.ai.

This is the help that comes with Pixaroma 1.4.212. Inside ComfyUI, select the node and click the orange ? above it to read the same help there.