This allows you to use Foveated Latent Sampling directly in ComfyUI. I developed this custom node to implement the method described in my article: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/22919">Foveated Latent Sampling - Article</a>.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:<ul><li>Implementation of Foveated Latent Sampling: Brings the theoretical method into a practical node.</li><li>Optimized Sampling: Focuses on coherence and detail based on the foveated rendering logic.</li><li>Easy Integration: Designed to work with standard latent handling in ComfyUI.</li></ul><hr /><h3 id="installation">📂 Installation</h3>Manual Installation:<ol><li>Download the <code>.zip</code> file.</li><li>Navigate to your ComfyUI <code>custom_nodes</code> folder (e.g., <code>ComfyUI/custom_nodes/</code>).</li><li>Extract the archive directly into this folder.<ul><li>Resulting path should look like: <code>ComfyUI/custom_nodes/Foveated-Latent-Sampling/</code></li></ul></li><li>Restart ComfyUI.</li></ol>Requirements: If the node relies on specific libraries not present in the default ComfyUI environment, run:<pre><code>pip install -r requirements.txt</code></pre>(Skip this if you don't encounter import errors upon restart).<hr /><h3 id="how-to-use">🚀 How to Use</h3><ol><li>After restarting ComfyUI, double-click on the canvas or right-click to open the node menu.</li><li>Search for "FLS" or look under the Sampler category.</li><li>Connect the node in your workflow (typically between the Latent source and the Sampler, or as a patch depending on the specific node logic).</li><li>Refer to the original <a target="_blank" rel="ugc" href="https://civarchive.com/articles/22919/bss-foveated-latent-sampling">Article</a> for theoretical background and parameter tuning tips.</li></ol><hr />Bug Reports &amp; 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.

Files with SHA256: 85cc747a01ac61e671fdde37380bfd7c5fbcb354f06a64f292d80951a8d79796