ComfyUI Node: Lora List Stacker

Lora List Stacker:

The AV_LoraListStacker node is designed to facilitate the loading and stacking of multiple LoRA (Low-Rank Adaptation) models in a sequential manner. This node is particularly useful for AI artists who want to apply multiple LoRA models to their base model and CLIP (Contrastive Language-Image Pre-Training) model, enhancing the creative possibilities and fine-tuning the output. By parsing a list of LoRA models, this node ensures that each model is loaded with specified strengths for both the model and the CLIP, allowing for a nuanced and layered application of LoRA effects. This node simplifies the process of managing multiple LoRA models, making it easier to experiment with different combinations and strengths to achieve the desired artistic results.

Lora List Stacker Input Parameters:

model

This parameter represents the base model to which the LoRA models will be applied. It is essential for the node's operation as it serves as the foundation upon which the LoRA models will be stacked.

clip

This parameter represents the CLIP model that will be used in conjunction with the base model. The CLIP model helps in understanding and processing the textual descriptions, enhancing the overall output when combined with the base model and LoRA models.

data

This parameter is a string that contains a list of LoRA models in JSON format. Each LoRA model in the list includes its name, strength for the model, and strength for the CLIP. The default value is an empty string, and it supports multiline input. This parameter is crucial as it dictates which LoRA models will be loaded and their respective strengths.

base_url

This optional parameter is a string that specifies the base URL for downloading LoRA models if they are not available locally. The default value is lora_cloud_front_url. This parameter is useful for ensuring that all required LoRA models are accessible, even if they need to be fetched from an external source.

Lora List Stacker Output Parameters:

MODEL

This output parameter represents the base model after all specified LoRA models have been applied. It reflects the cumulative effect of the stacked LoRA models, providing a modified version of the original base model.

CLIP

This output parameter represents the CLIP model after all specified LoRA models have been applied. Similar to the base model, it reflects the cumulative effect of the stacked LoRA models on the CLIP model, enhancing its ability to process and understand textual descriptions.

Lora List Stacker Usage Tips:

• Ensure that the data parameter is correctly formatted in JSON and includes all necessary LoRA models with their respective strengths.
• Use the base_url parameter to specify a reliable source for downloading LoRA models that are not available locally, ensuring that all required models are accessible.
• Experiment with different strengths for the model and CLIP to achieve the desired artistic effects, as varying these values can significantly impact the output.

Lora List Stacker Common Errors and Solutions:

Loading lora list: <data>

• Explanation: This message indicates that the node is attempting to load the list of LoRA models specified in the data parameter.
• Solution: Ensure that the data parameter is correctly formatted in JSON and includes valid LoRA model names and strengths.

Warning: Not found Lora model <lora_override>. Use <lora_name> instead.

• Explanation: This warning occurs when the specified LoRA model override is not found in the local directory.
• Solution: Verify that the lora_override parameter is set to a valid LoRA model name available in the local directory or ensure that the base_url is correctly set to download the required model.

AssertionError: preprocessor must be set when using Auto mode

• Explanation: This error occurs when the preprocessor is not set while using the Auto mode for control_net_name.
• Solution: Ensure that the preprocessor parameter is set to a valid value when using Auto mode for the control_net_name parameter.