This allows you to use Foveated Latent Sampling directly in ComfyUI. I developed this custom node to implement the method described in my article: Foveated Latent Sampling - Article.

This node optimizes the latent sampling process (as discussed in the article) to improve structure and detail distribution. It is packaged here as a standalone installation for easy integration into your workflows.

Key Features:

Implementation of Foveated Latent Sampling: Brings the theoretical method into a practical node.
Optimized Sampling: Focuses on coherence and detail based on the foveated rendering logic.
Easy Integration: Designed to work with standard latent handling in ComfyUI.

📂 Installation

Manual Installation:

Download the .zip file.
Navigate to your ComfyUI custom_nodes folder (e.g., ComfyUI/custom_nodes/).
Extract the archive directly into this folder.
- Resulting path should look like: ComfyUI/custom_nodes/Foveated-Latent-Sampling/
Restart ComfyUI.

Requirements: If the node relies on specific libraries not present in the default ComfyUI environment, run:

pip install -r requirements.txt

(Skip this if you don't encounter import errors upon restart).

🚀 How to Use

After restarting ComfyUI, double-click on the canvas or right-click to open the node menu.
Search for "FLS" or look under the Sampler category.
Connect the node in your workflow (typically between the Latent source and the Sampler, or as a patch depending on the specific node logic).
Refer to the original Article for theoretical background and parameter tuning tips.

Bug Reports & Feedback: This node is an experimental implementation of the FLS concept. If you encounter issues or have suggestions for the sampling logic, please leave a comment below.

Search AI Models & Addons

very few results