Importing Your Own LoRA Models
LoRA (Low-Rank Adaptation) models allow you to customize generative AI models with specific styles, characters, or concepts. Here's how to import your own LoRA models into Weavy:
Access the Import LoRA Node:
Open the toolbox menu in the top left corner and drag out the Import LoRA node
Alternatively, right-click anywhere on the canvas, type "LoRA" in the search box, and select the Import LoRA node
Upload Your LoRA File:
Click on the Import LoRA node
Click the "Upload" button
Select your LoRA file from your computer
The node will display the name of your imported LoRA when successfully uploaded
Connect to a Compatible Model:
Add a model that supports LoRAs, such as FluxDev LoRA (you can find this by typing "FluxDev LoRA" in the search box)
Connect your text prompt to the "prompt" input of the FluxDev LoRA model
Connect the Import LoRA node to the "LoRA1" input of the FluxDev LoRA model
Control LoRA Strength (optional):
To control the influence of your LoRA, add a Number node (found in datatypes or through the quick search menu)
For a full-strength effect, set the number to 1
For a more subtle effect, use a decimal value by:
Opening the "show more" button on the Number node
Changing the type to "float"
Setting a decimal value between 0 and 1 (e.g., 0.7)
Connect the Number node to the "LoRA1 scale" input on the FluxDev LoRA model
Using Multiple LoRAs:
You can add a second LoRA by importing another LoRA file with a new Import LoRA node
Connect this second LoRA to the "LoRA2" input of the FluxDev LoRA model
Control its strength with another Number node connected to "LoRA2 scale"
Your imported LoRA will be saved with your workflow, allowing you to reuse it in future sessions.
Tips for Working with Custom Models
Test Different Strength Values: When using LoRAs, experiment with different strength values to find the right balance between the LoRA's style and the base model.
Combine LoRAs: Try combining multiple LoRAs for unique effects, but be aware that not all combinations work well together.
Check Model Requirements: External models from Replicate and Civit AI may have specific input requirements. Check the model's documentation for guidance.
Save Successful Workflows: When you find a combination of models and settings that works well, save your workflow for future use.
Manage Your Library: Periodically organize your saved models by giving them clear names and descriptions.
By mastering the art of importing custom models and LoRAs, you'll significantly expand Weavy's capabilities and tailor the platform to your specific creative needs.