ComfyUI > Nodes > content_aware_tiles > Reject Candidate Tiles

ComfyUI Node: Reject Candidate Tiles

Class Name

RejectCandidateTiles

Category
tiles
Author
samsartor (Account age: 5064days)
Extension
content_aware_tiles
Latest Updated
2026-01-27
Github Stars
0.02K

How to Install content_aware_tiles

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

Reject Candidate Tiles Description

Optimizes image tile selection by evaluating candidates with metrics like sifid or textile.

Reject Candidate Tiles:

The RejectCandidateTiles node is designed to refine and optimize the selection of image tiles from a set of candidates based on specific evaluation metrics. This node is particularly useful in scenarios where you have multiple candidate tiles for a given position and need to select the best ones according to a chosen criterion. By leveraging metrics such as sifid, textile, or random, the node evaluates each candidate tile against a source image or other criteria, ensuring that only the most suitable tiles are retained. This process enhances the quality and coherence of the final image composition, making it a valuable tool for AI artists looking to create visually appealing and contextually relevant tile-based artworks.

Reject Candidate Tiles Input Parameters:

tiles

The tiles parameter represents the input image tiles that are candidates for selection. These tiles are evaluated based on the specified metric to determine which ones should be retained. The quality and characteristics of these tiles directly impact the final output, as they form the basis of the selection process.

tileset

The tileset parameter is a structured input that contains metadata about the tiles, such as their type and resolution. It is crucial for ensuring that the node processes the tiles correctly and maintains consistency in the output. The tileset must be provided as a forced input, ensuring that the node has all necessary information to function properly.

source

The source parameter is an image tensor used for comparison when evaluating candidate tiles with certain metrics, such as sifid. This source image acts as a reference point, allowing the node to assess how well each candidate tile matches or complements the source, thereby influencing the selection process.

metric

The metric parameter specifies the criterion used to evaluate and select the best candidate tiles. Available options include sifid, textile, and random. Each metric offers a different approach to tile evaluation: sifid compares tiles to the source image, textile uses a pre-trained model for evaluation, and random selects tiles without a specific criterion. The choice of metric significantly affects the selection outcome and should align with the desired artistic effect.

keep

The keep parameter determines the number of best tiles to retain per position after evaluation. It is an integer value with a default of 1, meaning that by default, only the top tile is kept. Adjusting this parameter allows for more or fewer tiles to be retained, providing flexibility in the final composition and enabling artists to control the level of detail and variation in their work.

Reject Candidate Tiles Output Parameters:

IMAGE

The IMAGE output is a tensor of the selected tiles that have been evaluated and retained based on the specified metric. This output represents the refined set of tiles that best meet the criteria set by the input parameters, ready for further processing or final composition.

TILESET

The TILESET output is an updated version of the input tileset, reflecting any changes made during the selection process. This includes adjustments to the number of candidates retained, ensuring that the tileset remains consistent with the selected tiles and is ready for subsequent use in the workflow.

Reject Candidate Tiles Usage Tips:

  • To achieve the best results, carefully choose the metric parameter based on the desired outcome. For instance, use sifid when you want tiles that closely match a source image, or textile for a more texture-oriented selection.
  • Adjust the keep parameter to control the diversity of the output. A higher value allows for more variation, which can be useful for creating complex patterns or textures.
  • Ensure that the tileset input is correctly configured to match the characteristics of the tiles, as this ensures proper processing and consistency in the output.

Reject Candidate Tiles Common Errors and Solutions:

Unsupported metric specified

  • Explanation: This error occurs when a metric that is not supported by the node is specified in the metric parameter.
  • Solution: Ensure that the metric parameter is set to one of the supported options: sifid, textile, or random.

Mismatch in candidates and keep values

  • Explanation: This error arises when the keep value exceeds the number of available candidates in the tileset.
  • Solution: Verify that the keep parameter is less than or equal to the number of candidates specified in the tileset. Adjust the keep value accordingly to avoid this error.

Reject Candidate Tiles Related Nodes

Go back to the extension to check out more related nodes.
content_aware_tiles
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.