LoRA Stacker Node Documentation

Overview

The LoRA Stacker is an advanced custom node within the ComfyUI framework, specifically designed to streamline workflows by allowing users to manage and utilize multiple LoRA (Low-Rank Adaptation) models simultaneously. This node facilitates the stacking of multiple LoRA models, enabling users to apply various model strengths and clip strengths efficiently.

Functionality

Purpose

The primary function of the LoRA Stacker node is to manage and stack multiple LoRA models, allowing for the application of these models with different strengths and settings. It simplifies workflows by consolidating multiple LoRA operations into a single node, thus minimizing the total node count and enhancing workflow efficiency.

Inputs

The LoRA Stacker node accepts the following inputs:

• Input Mode: Two modes are available: "simple" and "advanced". This dictates how LoRA weights are handled.

• LoRA Count: Specifies the number of LoRA models to stack. The value can range from 0 to 50.

• LoRA Name (1 to 50): The names of the LoRA models to be stacked. The node supports up to 50 named inputs.

• LoRA Weight (1 to 50): Only relevant in "simple" mode. Specifies the weight to apply uniformly to both model and clip for each LoRA.

• Model Strength (1 to 50): Only relevant in "advanced" mode. Specifies the strength at which the model part of the LoRA is applied.

• Clip Strength (1 to 50): Only relevant in "advanced" mode. Specifies the strength at which the clip part of the LoRA is applied.

• LoRA Stack (Optional): An existing stack of LoRA models, allowing for further stacking operations.

Outputs

The LoRA Stacker node produces the following output:

• LORA Stack: This is a stack of tuples where each tuple contains the LoRA model name, model strength, and clip strength. This stack can be used in subsequent nodes that accept a LORA stack input, enabling further manipulation and usage of the stacked LoRA models.

Use in Workflows

In ComfyUI workflows, the LoRA Stacker node is typically used to facilitate the combined use of multiple LoRA models. Users can dynamically configure the weights and strengths of multiple LoRA models to find the optimal settings for their specific tasks or creative outputs.

Example Usage

1. Simple Mode: You would use the "simple" mode when you want to apply a uniform weight to both the model and clip parts of the LoRA, without the need for separate configurations. You would enter the LoRA model names and a single weight value.

2. Advanced Mode: Opt for the "advanced" mode when you need granular control over each LoRA model's application settings. This would involve specifying each LoRA's model strength and clip strength separately.

3. Combining LoRA Stacks: If you already have a predefined stack of LoRA models and wish to add more or modify the existing ones, you can input your existing stack, and the node will augment or modify it based on new inputs.

Special Features and Considerations

• Scalability: The node is capable of managing up to 50 LoRA models, making it suitable for complex workflows that require extensive model stacking.

• Convenience: By consolidating multiple LoRA operations into a single node, it reduces the complexity of workflows, which can be especially beneficial in large-scale projects.

• Flexibility: The two modes (simple and advanced) provide flexibility for users based on their proficiency and specific requirements—simple mode for quick setups and advanced mode for detailed customizations.

Conclusion

The LoRA Stacker node is a vital tool for users working with multiple LoRA models in ComfyUI, offering a flexible and efficient way to manage model stacking. By providing both simple and advanced configuration options, it caters to both quick application scenarios and in-depth, customizable workflows.