Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > ComfyUI-LoraBlockWeight > LoRA Block Weight (SD3.5 Large)

ComfyUI Node: LoRA Block Weight (SD3.5 Large)

Class Name

LoraBlockWeightSD35

Category
LoraBlockWeight
Author
Baldwinzc (Account age: 2432days)
Extension
ComfyUI-LoraBlockWeight
Latest Updated
2026-06-09
Github Stars
0.02K

How to Install ComfyUI-LoraBlockWeight

Install this extension via the ComfyUI Manager by searching for ComfyUI-LoraBlockWeight
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI-LoraBlockWeight in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

LoRA Block Weight (SD3.5 Large) Description

Facilitates LoRA block weight adjustments for Stable Diffusion 3.5 Large model's attention mechanisms.

LoRA Block Weight (SD3.5 Large):

The LoraBlockWeightSD35 node is designed to facilitate the application of LoRA (Low-Rank Adaptation) block weights specifically for the Stable Diffusion 3.5 Large model. This node is part of a suite of tools that allow you to fine-tune the attention mechanisms within the model by adjusting the weights of specific blocks, known as joint_blocks. These blocks are integral to the model's ability to process and integrate image and text data simultaneously. By leveraging the LoraBlockWeightSD35, you can enhance the model's performance on specific tasks by customizing how different parts of the model are weighted during inference. This capability is particularly beneficial for AI artists looking to achieve more nuanced and tailored outputs from their models, as it allows for precise control over the model's behavior without the need for extensive retraining.

LoRA Block Weight (SD3.5 Large) Input Parameters:

model

The model parameter specifies the Stable Diffusion 3.5 Large model to which the LoRA block weights will be applied. This parameter is crucial as it determines the specific architecture and block configuration that the node will interact with. There are no explicit minimum, maximum, or default values for this parameter, as it is dependent on the model being used.

vae

The VAE (Variational Autoencoder) parameter is used to define the VAE model that will be utilized in conjunction with the Stable Diffusion model. This parameter impacts the quality and style of the generated images, as the VAE is responsible for encoding and decoding image data. Like the model parameter, there are no specific value constraints, but it should be compatible with the chosen Stable Diffusion model.

lora_name

This parameter specifies the name of the LoRA configuration to be applied. It is essential for identifying which set of LoRA weights should be used during the block weight adjustment process. The lora_name must correspond to a valid LoRA configuration available in your environment.

positive

The positive parameter is used to define the positive prompt or input that guides the model's output. This input influences the model's focus and the elements it emphasizes in the generated image. There are no strict value constraints, but it should be a meaningful prompt relevant to the desired output.

negative

The negative parameter serves as the negative prompt or input, which helps the model understand what elements to de-emphasize or avoid in the output. This parameter is useful for refining the model's focus and ensuring that unwanted features are minimized. Like the positive parameter, it should be a relevant and meaningful prompt.

latent_image

This parameter represents the latent image data that the model will process. It is a crucial input as it forms the basis of the image generation process. The latent_image should be compatible with the model's input requirements, but specific value constraints are not provided.

seed

The seed parameter is used to initialize the random number generator, ensuring reproducibility of results. By setting a specific seed value, you can achieve consistent outputs across different runs. The seed can be any integer value, with no specific minimum or maximum constraints.

steps

This parameter defines the number of inference steps the model will perform. It directly impacts the quality and detail of the generated image, with more steps generally leading to higher quality outputs. The minimum value is typically 1, with no strict maximum, but practical limits depend on computational resources.

cfg

The cfg (Classifier-Free Guidance) parameter controls the strength of the guidance applied during inference. It influences how closely the model adheres to the provided prompts. The cfg value is typically a floating-point number, with higher values leading to stronger adherence to the prompts.

sampler_name

This parameter specifies the name of the sampling method to be used during inference. Different samplers can affect the style and quality of the generated images. The sampler_name should correspond to a valid sampling method available in your environment.

scheduler

The scheduler parameter defines the scheduling strategy for the inference process. It impacts how the model's weights are adjusted over the course of the inference steps. The scheduler should be compatible with the chosen model and sampling method.

denoise

This parameter controls the level of denoising applied to the generated image. It affects the clarity and smoothness of the output, with higher values typically leading to cleaner images. The denoise value is usually a floating-point number, with no strict constraints.

block_list

The block_list parameter specifies which blocks within the model should have their weights adjusted. It allows for targeted fine-tuning of specific parts of the model. The block_list should be a list of valid block identifiers, such as J00, J01, etc.

value_list

This parameter defines the specific weight values to be applied to the blocks specified in the block_list. It allows for precise control over the model's behavior by adjusting the influence of each block. The value_list should correspond to the block_list in length and order.

baseline_weight

The baseline_weight parameter sets the default weight value for blocks not explicitly listed in the block_list. It provides a reference point for the weight adjustments and ensures consistent behavior across the model. The baseline_weight is typically a floating-point number.

LoRA Block Weight (SD3.5 Large) Output Parameters:

adjusted_model

The adjusted_model output represents the Stable Diffusion model with the applied LoRA block weight adjustments. This output is crucial as it reflects the customized model configuration that can be used for generating images with the desired characteristics. The adjusted_model retains the original model's architecture but with modified block weights to achieve specific artistic effects.

LoRA Block Weight (SD3.5 Large) Usage Tips:

  • Experiment with different block_list and value_list combinations to achieve unique artistic styles and effects in your generated images.
  • Use the seed parameter to ensure reproducibility of results, allowing you to fine-tune your configurations and achieve consistent outputs.
  • Adjust the cfg parameter to balance between adhering to prompts and allowing for creative variations in the output.

LoRA Block Weight (SD3.5 Large) Common Errors and Solutions:

Invalid block identifier in block_list

  • Explanation: This error occurs when the block_list contains identifiers that do not match the model's block configuration.
  • Solution: Ensure that all block identifiers in the block_list are valid and correspond to the model's joint_blocks, such as J00, J01, etc.

Mismatched block_list and value_list lengths

  • Explanation: This error arises when the block_list and value_list have different lengths, leading to misalignment in weight assignments.
  • Solution: Verify that the block_list and value_list have the same number of elements, ensuring each block has a corresponding weight value.

Incompatible VAE model

  • Explanation: This error occurs when the specified VAE model is not compatible with the Stable Diffusion model.
  • Solution: Select a VAE model that is compatible with the Stable Diffusion 3.5 Large model to ensure proper encoding and decoding of image data.

LoRA Block Weight (SD3.5 Large) Related Nodes

Go back to the extension to check out more related nodes.
ComfyUI-LoraBlockWeight
RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

LoRA Block Weight (SD3.5 Large)