ComfyUI > Nodes > ComfyUI-ArchAi3d-Qwen > 🚀 Smart USDU Split-Latent

ComfyUI Node: 🚀 Smart USDU Split-Latent

Class Name

ArchAi3D_Smart_USDU_Split_Latent

Category
ArchAi3d/Upscaling/USDU
Author
Amir Ferdos (ArchAi3d) (Account age: 1109days)
Extension
ComfyUI-ArchAi3d-Qwen
Latest Updated
2026-04-17
Github Stars
0.05K

How to Install ComfyUI-ArchAi3d-Qwen

Install this extension via the ComfyUI Manager by searching for ComfyUI-ArchAi3d-Qwen
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI-ArchAi3d-Qwen 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

🚀 Smart USDU Split-Latent Description

Enhances latent image processing with split-latent workflow, mask cropping, and dual noise injection.

🚀 Smart USDU Split-Latent:

The ArchAi3D_Smart_USDU_Split_Latent node is designed to enhance the processing of latent images by implementing a sophisticated split-latent workflow. This node replaces the standard image processing logic with a more advanced approach that includes per-tile mask cropping, dual noise injection, and mask-based latent blending. It also supports per-tile conditioning, which allows for more precise and controlled image generation. The primary goal of this node is to provide a more refined and detailed output by leveraging these advanced techniques, making it particularly beneficial for AI artists looking to achieve high-quality results in their projects.

🚀 Smart USDU Split-Latent Input Parameters:

shared.batch

This parameter represents the batch of images to be processed. It is crucial for determining the number of images that will undergo the split-latent processing workflow. The batch size can significantly impact the processing time and resource usage.

model

The model parameter specifies the AI model used for processing the images. This model is responsible for interpreting the latent space and generating the final output. Choosing the right model can affect the style and quality of the generated images.

conditionings

Conditionings are used to guide the image generation process. They provide additional context or constraints that the model uses to produce more targeted results. This parameter is essential for achieving specific artistic effects or adhering to particular themes.

negative

The negative parameter is used to specify any negative conditions or constraints that should be avoided during image generation. It helps in refining the output by excluding unwanted elements or features.

vae

The VAE (Variational Autoencoder) parameter is involved in encoding and decoding the latent space. It plays a critical role in ensuring that the latent representations are accurately transformed into the final image outputs.

seed

The seed parameter is used to initialize the random number generator, ensuring reproducibility of results. By setting a specific seed, you can achieve consistent outputs across multiple runs.

steps

This parameter defines the number of steps the model will take during the image generation process. More steps can lead to higher quality images but may also increase processing time.

cfg

CFG (Classifier-Free Guidance) is a parameter that influences the balance between adhering to the conditioning and exploring new possibilities. Adjusting this parameter can affect the creativity and adherence to the specified conditions.

sampler_name

The sampler_name parameter specifies the sampling method used during image generation. Different samplers can produce varying results in terms of style and quality.

scheduler

The scheduler parameter determines the scheduling strategy for the image generation process. It can impact the efficiency and speed of the processing workflow.

denoise_high

This parameter sets the level of noise reduction applied during the high denoise phase. It is crucial for achieving a clean and polished final output.

upscale_by

The upscale_by parameter specifies the factor by which the image resolution is increased. Upscaling can enhance the detail and clarity of the final image.

force_uniform_tiles

This parameter ensures that all tiles are processed uniformly, which can be important for maintaining consistency across the entire image.

tiled_decode

Tiled_decode is a parameter that enables the decoding of images in tiles, which can improve processing efficiency and manage memory usage effectively.

tile_width

The tile_width parameter defines the width of each tile used in the split-latent processing. It can affect the granularity and detail of the final output.

tile_height

Similar to tile_width, the tile_height parameter specifies the height of each tile. Together, these parameters determine the size and aspect ratio of the tiles.

MODES[mode_type]

This parameter selects the mode of operation for the split-latent processing. Different modes can offer various processing techniques and effects.

SEAM_FIX_MODES[seam_fix_mode]

The seam_fix_mode parameter addresses any seams or artifacts that may occur between tiles. It ensures a seamless and cohesive final image.

mask_pil

The mask_pil parameter provides a mask image that guides the blending and processing of the latent space. It is essential for achieving specific artistic effects or focusing on particular areas of the image.

denoise_high_val

This parameter sets the specific value for high-level noise reduction, impacting the clarity and smoothness of the final output.

denoise_low_val

The denoise_low_val parameter specifies the value for low-level noise reduction, allowing for finer control over the noise levels in the image.

🚀 Smart USDU Split-Latent Output Parameters:

processed_images

The processed_images output parameter contains the final images generated after applying the split-latent processing workflow. These images are the result of the advanced techniques implemented by the node, offering high-quality and detailed outputs suitable for various artistic applications.

🚀 Smart USDU Split-Latent Usage Tips:

  • Experiment with different models and conditionings to achieve diverse artistic styles and effects.
  • Adjust the denoise_high and denoise_low parameters to find the right balance between detail and smoothness in your images.
  • Utilize the upscale_by parameter to enhance the resolution of your final outputs, especially for large-scale projects.

🚀 Smart USDU Split-Latent Common Errors and Solutions:

"Invalid model specified"

  • Explanation: This error occurs when the model parameter is not set correctly or the specified model is unavailable.
  • Solution: Ensure that the model parameter is set to a valid and available model. Check the documentation for supported models.

"Batch size exceeds limit"

  • Explanation: The batch size specified in shared.batch is too large for the available resources.
  • Solution: Reduce the batch size to a manageable number that fits within your system's capabilities.

"Tile dimensions are invalid"

  • Explanation: The specified tile_width or tile_height parameters are not valid.
  • Solution: Verify that the tile dimensions are set to reasonable values that align with the overall image size and aspect ratio.

🚀 Smart USDU Split-Latent Related Nodes

Go back to the extension to check out more related nodes.
ComfyUI-ArchAi3d-Qwen
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.