ComfyUI Node: 💊 CR Random LoRA Stack

💊 CR Random LoRA Stack:

The CR Random LoRA Stack node is designed to manage and apply multiple LoRA (Low-Rank Adaptation) models in a randomized manner, enhancing the flexibility and creativity of AI-generated art. This node allows you to stack up to three different LoRA models and apply them either exclusively or in combination, based on specified probabilities. The node supports features like deduplication of LoRA names, stride-based randomization, and forced re-randomization to ensure diverse and unique outputs. By leveraging this node, you can introduce variability and control over the application of LoRA models, making it a powerful tool for AI artists looking to experiment with different styles and effects.

💊 CR Random LoRA Stack Input Parameters:

exclusive_mode

This parameter determines whether only one LoRA model should be applied exclusively. When set to "On," the node evaluates the chances of each LoRA model being applied and selects the one with the highest probability. This ensures that only one LoRA model is active at a time, providing a clear and distinct effect. Options: "On", "Off".

stride

The stride parameter sets the minimum number of cycles before a re-randomization of the LoRA models is performed. This helps in maintaining consistency over a specified number of iterations before introducing variability again. Minimum value: 1, Maximum value: N (where N is the number of cycles you want), Default value: 1.

force_randomize_after_stride

When this parameter is set to "On," it forces the node to re-randomize the LoRA models after the specified stride, even if the same set of models was selected previously. This ensures that the output remains varied and prevents repetitive patterns. Options: "On", "Off".

lora_name_1

The name of the first LoRA model to be considered for stacking. This parameter should be set to the specific identifier of the LoRA model you wish to use. Default value: "None".

model_weight_1

The weight assigned to the first LoRA model, determining its influence on the final output. This value should be set based on the desired impact of the model. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

clip_weight_1

The weight assigned to the CLIP (Contrastive Language-Image Pre-Training) model associated with the first LoRA model. This value affects the text-to-image alignment. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

switch_1

This parameter controls whether the first LoRA model is active. When set to "On," the model is considered for stacking. Options: "On", "Off".

chance_1

The probability of the first LoRA model being applied when exclusive mode is off. This value should be set based on the desired likelihood of the model's application. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

lora_name_2

The name of the second LoRA model to be considered for stacking. This parameter should be set to the specific identifier of the LoRA model you wish to use. Default value: "None".

model_weight_2

The weight assigned to the second LoRA model, determining its influence on the final output. This value should be set based on the desired impact of the model. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

clip_weight_2

The weight assigned to the CLIP model associated with the second LoRA model. This value affects the text-to-image alignment. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

switch_2

This parameter controls whether the second LoRA model is active. When set to "On," the model is considered for stacking. Options: "On", "Off".

chance_2

The probability of the second LoRA model being applied when exclusive mode is off. This value should be set based on the desired likelihood of the model's application. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

lora_name_3

The name of the third LoRA model to be considered for stacking. This parameter should be set to the specific identifier of the LoRA model you wish to use. Default value: "None".

model_weight_3

The weight assigned to the third LoRA model, determining its influence on the final output. This value should be set based on the desired impact of the model. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

clip_weight_3

The weight assigned to the CLIP model associated with the third LoRA model. This value affects the text-to-image alignment. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

switch_3

This parameter controls whether the third LoRA model is active. When set to "On," the model is considered for stacking. Options: "On", "Off".

chance_3

The probability of the third LoRA model being applied when exclusive mode is off. This value should be set based on the desired likelihood of the model's application. Minimum value: 0.0, Maximum value: 1.0, Default value: 0.5.

lora_stack

An optional parameter that allows you to provide an existing stack of LoRA models. This stack will be extended with the new models based on the current configuration. Default value: None.

💊 CR Random LoRA Stack Output Parameters:

LORA_STACK

The output is a list of tuples representing the stacked LoRA models. Each tuple contains the name of the LoRA model, its model weight, and its CLIP weight. This stack can be used in subsequent nodes to apply the combined effects of the selected LoRA models, providing a versatile and dynamic approach to AI-generated art.

💊 CR Random LoRA Stack Usage Tips:

• To ensure a diverse range of outputs, experiment with different stride values and enable the force_randomize_after_stride parameter.
• Use exclusive_mode to focus on the distinct effects of individual LoRA models, especially when you want to highlight specific styles or features.
• Adjust the model_weight and clip_weight parameters to fine-tune the influence of each LoRA model on the final output, balancing their contributions as needed.

💊 CR Random LoRA Stack Common Errors and Solutions:

"LoRA model name not found"

• Explanation: This error occurs when the specified LoRA model name does not exist or is incorrectly spelled.
• Solution: Double-check the LoRA model names and ensure they are correctly specified.

"Invalid weight value"

• Explanation: This error occurs when the model_weight or clip_weight parameters are set outside the valid range of 0.0 to 1.0.
• Solution: Ensure that the weight values are within the valid range and adjust them accordingly.

"Stride value must be greater than 0"

• Explanation: This error occurs when the stride parameter is set to a value less than 1. - Solution: Set the stride parameter to a value of 1 or higher to ensure proper functionality.

"No active LoRA models"

• Explanation: This error occurs when none of the LoRA models are set to "On" or all have a chance of 0.0.
• Solution: Ensure that at least one LoRA model is active and has a non-zero chance of being applied.