ComfyUI  >  Nodes  >  Tiled sampling for ComfyUI >  Tiled KSampler

ComfyUI Node: Tiled KSampler

Class Name

BNK_TiledKSampler

Category
sampling
Author
BlenderNeko (Account age: 532 days)
Extension
Tiled sampling for ComfyUI
Latest Updated
5/22/2024
Github Stars
0.3K

How to Install Tiled sampling for ComfyUI

Install this extension via the ComfyUI Manager by searching for  Tiled sampling for ComfyUI
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter Tiled sampling for ComfyUI 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
  • High-speed GPU machines
  • 200+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 50+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

Tiled KSampler Description

Facilitates high-quality image generation through tiled sampling for detailed and refined results, with support for various tiling strategies and advanced features.

Tiled KSampler:

The BNK_TiledKSampler node is designed to facilitate the generation of high-quality images by leveraging a tiled sampling approach. This method is particularly beneficial for creating large images or images with intricate details, as it processes smaller sections (tiles) of the image sequentially. By doing so, it ensures that each tile receives focused attention, leading to more refined and coherent results. The node integrates various strategies for tiling, such as "padded" and "random strict," to accommodate different artistic needs and preferences. Additionally, it supports advanced features like conditional sampling and denoising, making it a versatile tool for AI artists aiming to enhance their creative workflows.

Tiled KSampler Input Parameters:

model

The model parameter specifies the AI model to be used for image generation. This model is responsible for interpreting the input conditions and generating the corresponding image tiles. The choice of model can significantly impact the style and quality of the output.

seed

The seed parameter sets the initial state for the random number generator used in the sampling process. By using the same seed, you can reproduce the same image output, which is useful for consistency and experimentation. The seed value can be any integer.

tile_width

The tile_width parameter defines the width of each tile in pixels. This determines how the image is divided horizontally for processing. Adjusting the tile width can affect the level of detail and the processing time. Typical values range from 64 to 512 pixels.

tile_height

The tile_height parameter defines the height of each tile in pixels. Similar to tile_width, this parameter controls the vertical division of the image. The choice of tile height can influence the image's detail and the efficiency of the sampling process. Typical values range from 64 to 512 pixels.

tiling_strategy

The tiling_strategy parameter specifies the method used to handle the edges and overlaps of the tiles. Options include "padded" and "random strict," each offering different approaches to ensure seamless transitions between tiles. The choice of strategy can affect the overall coherence of the final image.

steps

The steps parameter determines the number of iterations the sampler will perform for each tile. More steps generally lead to higher quality images but also increase the processing time. Typical values range from 10 to 1000 steps.

cfg

The cfg (Classifier-Free Guidance) parameter controls the strength of the guidance applied during sampling. Higher values result in images that more closely follow the input conditions, while lower values allow for more creative freedom. Typical values range from 1.0 to 20.0.

sampler_name

The sampler_name parameter specifies the type of sampler to be used, such as "Euler" or "LMS." Different samplers can produce varying styles and qualities of images, so experimenting with different options can yield diverse results.

scheduler

The scheduler parameter defines the scheduling strategy for the sampling process. This can influence the progression and refinement of the image over the sampling steps. Common options include "linear" and "cosine."

positive

The positive parameter is a list of conditions that positively influence the image generation. These conditions guide the model towards desired features and elements in the final image.

negative

The negative parameter is a list of conditions that negatively influence the image generation. These conditions help the model avoid unwanted features and elements in the final image.

latent_image

The latent_image parameter provides an initial latent representation of the image, which can be refined through the sampling process. This can be useful for starting from a specific base image or latent state.

denoise

The denoise parameter controls the amount of denoising applied during the sampling process. Higher values result in smoother images, while lower values retain more detail and texture. Typical values range from 0.1 to 1.0.

Tiled KSampler Output Parameters:

sampled_image

The sampled_image parameter is the final output image generated by the node. This image is the result of the tiled sampling process, incorporating all the specified conditions and parameters. It represents the culmination of the model's interpretation and refinement of the input data.

Tiled KSampler Usage Tips:

  • Experiment with different tiling strategies to find the one that best suits your artistic needs. "Padded" can help with seamless transitions, while "random strict" can add interesting variations.
  • Adjust the tile width and height based on the level of detail you want in your image. Smaller tiles can capture finer details but may increase processing time.
  • Use the seed parameter to reproduce specific results, which is useful for iterative improvements and consistency in your projects.
  • Vary the steps parameter to balance between image quality and processing time. More steps generally yield better results but require more computational resources.

Tiled KSampler Common Errors and Solutions:

"Invalid tile dimensions"

  • Explanation: The specified tile width or height is not within the acceptable range.
  • Solution: Ensure that the tile_width and tile_height parameters are set to values between 64 and 512 pixels.

"Model not found"

  • Explanation: The specified model is not available or incorrectly referenced.
  • Solution: Verify that the model parameter is correctly set to an available AI model.

"Incompatible tiling strategy"

  • Explanation: The chosen tiling strategy is not supported by the current configuration.
  • Solution: Check the tiling_strategy parameter and select a valid option such as "padded" or "random strict."

"Denoise value out of range"

  • Explanation: The denoise parameter is set to a value outside the acceptable range.
  • Solution: Adjust the denoise parameter to a value between 0.1 and 1.0.

Tiled KSampler Related Nodes

Go back to the extension to check out more related nodes.
Tiled sampling for ComfyUI
RunComfy

© Copyright 2024 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals.