Visit ComfyUI Online for ready-to-use ComfyUI environment
Enhance AI-generated images with advanced multiscale control for refined results.
FreeU (Advanced) is a sophisticated node designed to enhance the quality and flexibility of AI-generated images by leveraging advanced multiscale techniques. This node allows you to fine-tune various parameters to achieve the desired level of detail and smoothness in different parts of the image. By adjusting the multiscale strength and selecting specific target blocks, you can control how the model processes different regions, leading to more refined and visually appealing results. The primary goal of FreeU (Advanced) is to provide AI artists with greater control over the image generation process, enabling the creation of high-quality, customized artwork.
This parameter specifies the AI model to be used for image generation. It is essential for defining the base model that will be manipulated by the FreeU (Advanced) node.
This parameter allows you to select which part of the model to target for adjustments. Options include "output_block", "middle_block", "input_block", and "all". By choosing a specific block, you can focus the enhancements on particular stages of the image generation process.
This parameter determines the mode of multiscale processing to be applied. The available options are defined by the mscales
dictionary. Selecting the appropriate mode can significantly impact the final image quality by adjusting how different scales are handled.
This parameter controls the strength of the multiscale effect, with a default value of 1.0. The range is from 0 to 1.0, with a step of 0.001. Adjusting this value allows you to fine-tune the intensity of the multiscale processing, balancing between detail and smoothness.
This parameter sets the size of the first slice for multiscale processing, with a default value of 640. The range is from 64 to 1280, with a step of 1. This parameter helps in defining the granularity of the multiscale slices.
This parameter sets the size of the second slice for multiscale processing, with a default value of 320. The range is from 64 to 640, with a step of 1. Similar to slice_b1
, this parameter helps in defining the granularity of the multiscale slices.
This parameter controls the first scaling factor for the multiscale processing, with a default value of 1.1. The range is from 0.0 to 10.0, with a step of 0.001. Adjusting this value influences the scaling applied to the first slice.
This parameter controls the second scaling factor for the multiscale processing, with a default value of 1.2. The range is from 0.0 to 10.0, with a step of 0.001. Adjusting this value influences the scaling applied to the second slice.
This parameter sets the first smoothing factor for the multiscale processing, with a default value of 0.9. The range is from 0.0 to 10.0, with a step of 0.001. This parameter helps in controlling the smoothness of the first slice.
This parameter sets the second smoothing factor for the multiscale processing, with a default value of 0.2. The range is from 0.0 to 10.0, with a step of 0.001. This parameter helps in controlling the smoothness of the second slice.
This output parameter represents the final processed image after applying the multiscale techniques. The result is a refined image with enhanced details and smoothness, based on the input parameters and the selected model.
target_block
settings to see how focusing on different parts of the model affects the final image.multiscale_strength
to find the right balance between detail and smoothness for your specific use case.slice_b1
and slice_b2
to control the granularity of the multiscale slices, which can help in achieving more precise enhancements.b1
, b2
, s1
, and s2
to customize the scaling and smoothing effects, allowing for a high degree of control over the image generation process.slice_b1
or slice_b2
) are set outside the allowed range.slice_b1
and 64 to 640 for slice_b2
).model
parameter is not provided.model
parameter before running the node.© Copyright 2024 RunComfy. All Rights Reserved.