Flux Model Merging

If you’ve been experimenting with Flux models and want to combine them for better performance or results, you're in the right place. Merging Flux models can significantly enhance your workflow, and with ComfyUI's latest update, it's easier than ever. Below, we’ll guide you through the process of merging Flux models and how to choose the right combination for your project.

Why Combine Flux Models?

When working with different models like Dev, Schnell, or others, you may want to merge them for improved performance or quality. Here’s a simple breakdown of the combinations:

• Dev with Dev: This is the classic approach where you create .safetensors files from the Dev model and run them as normal checkpoints. This ensures stable results but may not always be the fastest.
• Dev with Schnell: This combination helps to speed up the loading process while maintaining an acceptable quality level. It’s a good choice if you need faster performance but can tolerate slight quality compromises.

Workflow Overview

Steps to Merge Flux Models

Follow these simple steps to merge your Flux models using ComfyU

Step 1: Upgrade to the Latest Version of ComfyUI

To begin, make sure your ComfyUI is updated to the latest version. The new update includes a handy button called ModelMergeFlux1, which is essential for merging Flux models.

Step 2: Create the Merge Button

Once your ComfyUI is up to date, you will see a new button called ModelMergeFlux1. Click on this button, and you’ll be prompted to choose the models you wish to merge. Select the following:

• VAE (Variational Autoencoder)
• UNET
• Flux

If you prefer, you can download pre-configured settings or models to simplify the process further.

Step 3: Merge the Models

After selecting your desired models (VAE, UNET, and Flux), click Queue. The system will begin merging the models into a .safetensors file. This process may take a little time, so be patient and let ComfyUI handle the merging.

Typically, it will be located at: ComfyUI/output/checkpoint. On MimicPC, its path is File > Storage > output > checkpoint.

Step 4: Load Your Workflow

Once the merge is complete, you can load the newly created .safetensors model into your workflow just like you would with any regular checkpoint. If you'd like, you can rename the merged model to something more meaningful, as the default name will be 00001.safetensors. After renaming it, move the model to the following directory: Storage > models > checkpoints. Once the model is in the right place, you can create and run your workflow as usual, enjoying the enhanced results with your merged Flux model.