ComfyUI Node: 💊 CR LoRA Stack

💊 CR LoRA Stack:

The CR LoRA Stack node is designed to facilitate the combination and application of multiple LoRA (Low-Rank Adaptation) models to enhance your AI art generation process. This node allows you to stack up to three different LoRA models, each with customizable weights for both the model and the CLIP (Contrastive Language-Image Pre-Training) components. By leveraging this node, you can create complex and nuanced modifications to your base model, enabling more sophisticated and varied artistic outputs. The primary goal of the CR LoRA Stack is to provide a flexible and powerful tool for AI artists to experiment with different LoRA combinations, thereby expanding the creative possibilities and fine-tuning the generated art to meet specific artistic visions.

💊 CR LoRA Stack Input Parameters:

lora_name_1

This parameter specifies the name of the first LoRA model to be included in the stack. It is essential for identifying which LoRA model to apply. If set to "None," this LoRA model will not be included in the stack. The impact of this parameter is significant as it determines the first layer of adaptation applied to your base model. There are no minimum or maximum values, but valid LoRA model names should be used.

model_weight_1

This parameter sets the weight for the first LoRA model's impact on the base model. It controls how strongly the first LoRA model influences the final output. The weight can range from 0.0 (no influence) to 1.0 (full influence), with a default value typically around 0.5 for balanced adaptation.

clip_weight_1

This parameter sets the weight for the first LoRA model's impact on the CLIP component. Similar to model_weight_1, it controls the influence on the CLIP model, ranging from 0.0 to 1.0, with a default value around 0.5.

switch_1

This parameter is a toggle switch that determines whether the first LoRA model is active ("On") or inactive ("Off"). When set to "Off," the first LoRA model is ignored, regardless of the other parameters.

lora_name_2

This parameter specifies the name of the second LoRA model to be included in the stack. It functions similarly to lora_name_1 and is crucial for adding another layer of adaptation. Valid LoRA model names should be used.

model_weight_2

This parameter sets the weight for the second LoRA model's impact on the base model. It ranges from 0.0 to 1.0, with a default value around 0.5, controlling the influence of the second LoRA model.

clip_weight_2

This parameter sets the weight for the second LoRA model's impact on the CLIP component. It ranges from 0.0 to 1.0, with a default value around 0.5, controlling the influence on the CLIP model.

switch_2

This parameter is a toggle switch that determines whether the second LoRA model is active ("On") or inactive ("Off"). When set to "Off," the second LoRA model is ignored.

lora_name_3

This parameter specifies the name of the third LoRA model to be included in the stack. It functions similarly to lora_name_1 and lora_name_2, adding another layer of adaptation. Valid LoRA model names should be used.

model_weight_3

This parameter sets the weight for the third LoRA model's impact on the base model. It ranges from 0.0 to 1.0, with a default value around 0.5, controlling the influence of the third LoRA model.

clip_weight_3

This parameter sets the weight for the third LoRA model's impact on the CLIP component. It ranges from 0.0 to 1.0, with a default value around 0.5, controlling the influence on the CLIP model.

switch_3

This parameter is a toggle switch that determines whether the third LoRA model is active ("On") or inactive ("Off"). When set to "Off," the third LoRA model is ignored.

lora_stack

This optional parameter allows you to pass an existing stack of LoRA models. If provided, the node will extend this stack with the specified LoRA models and their respective weights. This parameter is useful for building upon previously defined stacks.

💊 CR LoRA Stack Output Parameters:

LORA_STACK

This output parameter returns the final stack of LoRA models, including all specified models and their respective weights. The stack is a list of tuples, where each tuple contains the LoRA model name, model weight, and CLIP weight. This output is crucial for further processing or applying the stacked LoRA models to your base model.

show_help

This output parameter provides a URL to the documentation or help page for the CR LoRA Stack node. It is useful for users who need additional information or guidance on using the node effectively.

💊 CR LoRA Stack Usage Tips:

• Ensure that the LoRA model names you specify are valid and available in your environment to avoid errors.
• Experiment with different weights for the model and CLIP components to achieve the desired artistic effect. Start with balanced weights (e.g., 0.5) and adjust as needed.
• Use the lora_stack parameter to build upon existing stacks, allowing for more complex and layered adaptations.
• Toggle the switches (switch_1, switch_2, switch_3) to quickly enable or disable specific LoRA models without removing them from the stack.

💊 CR LoRA Stack Common Errors and Solutions:

Invalid LoRA model name

• Explanation: The specified LoRA model name does not exist or is not valid.
• Solution: Ensure that the LoRA model names you provide are correct and available in your environment.

Weight out of range

• Explanation: The specified weight for the model or CLIP component is outside the acceptable range (0.0 to 1.0).
• Solution: Adjust the weight values to be within the range of 0.0 to 1.0.

LoRA model not applied

• Explanation: The switch for the LoRA model is set to "Off," so the model is not included in the stack.
• Solution: Set the switch to "On" to include the LoRA model in the stack.

Empty LoRA stack

• Explanation: No LoRA models are specified or all switches are set to "Off," resulting in an empty stack.
• Solution: Ensure that at least one LoRA model is specified and its switch is set to "On."