ComfyUI-spritefusion-pixel-snapper

Sprite Fusion Pixel Snapper for ComfyUI

Sprite Fusion Pixel Snapper for ComfyUI icon

This custom node is a Python port of the Sprite Fusion Pixel Snapper tool that fixes messy AI-generated pixel art by snapping it to a clean grid and quantized palette.

Install

Place this folder inside ComfyUI/custom_nodes/ and restart ComfyUI.

Node

• Name: Sprite Fusion Pixel Snapper
• Category: image/transform

Inputs

• image – ComfyUI IMAGE tensor (batch supported; returns a list).
• k_colors (int) – Palette size (default 16).
• k_seed (int) – RNG seed for palette init (default 42).
• pixel_size (float) – Override the auto-detected input pixel size; 0.0 keeps auto-detection. Values must be between 1 and half the smallest image dimension.
• output_scale (int) – Optional integer upscaling after snapping (nearest-neighbor, default 1, max 16).

Additional tuning parameters: max_kmeans_iterations, peak_threshold_multiplier, peak_distance_filter, walker_search_window_ratio, walker_min_search_window, walker_strength_threshold, min_cuts_per_axis, fallback_target_segments, max_step_ratio.

Output

• IMAGE (List) – per-frame outputs as a list, preserving each frame's size.

Compatibility note

This is a Python port, not a byte-for-byte Rust wrapper; results may differ from upstream.

Credits

• Upstream repository: https://github.com/Hugo-Dz/spritefusion-pixel-snapper

License

• This repository: MIT (see LICENSE, © 2025 x0x0b).
• Upstream repository: MIT (see LICENSE-spritefusion-pixel-snapper, © 2025 Hugo Duprez).