2🐕Conventional filters

There's a specific detail worth knowing about this node before you even open the dropdown: the filter names it exposes - BLUR, CONTOUR, DETAIL, EDGE_ENHANCE, EDGE_ENHANCE_MORE, EMBOSS, FIND_EDGES, GaussianBlur, MaxFilter, MedianFilter, MinFilter, ModeFilter - aren't ergouzi's own invention. They're a direct, verbatim match for Python Pillow's built-in ImageFilter module. This node is a thin, honest wrapper around Pillow's classic filter kernels, exposed as a single dropdown instead of you writing a script. That's not a knock - it means the behavior is exactly as predictable and well-documented as Pillow itself, and none of it needs a model or GPU pass; these are fast, deterministic kernel operations.

How it works

Two required inputs: image, and filter_type, a dropdown of 16 choices (the schema samples twelve of them, listed above, with four more rounding out the set - almost certainly the rest of Pillow's stock filter kit, things like SHARPEN and SMOOTH). Pick one, run it, get back a single filtered IMAGE. No strength or radius controls beyond what's baked into each preset - this isn't a tunable Gaussian blur with an adjustable sigma, it's the fixed Pillow default for whichever filter you selected (with the partial exception of GaussianBlur, which even in Pillow ships with its own default radius unless a wrapper exposes the parameter - this node's schema shows no separate radius input, so expect the library default).

What each type is actually useful for

You don't need to memorize all sixteen - a handful cover most real use: GaussianBlur/BLUR for softening, DETAIL and EDGE_ENHANCE/EDGE_ENHANCE_MORE for sharpening emphasis, FIND_EDGES/CONTOUR for line-art/edge extraction (useful upstream of a ControlNet-style workflow if you need a quick edge map), EMBOSS for a stylized relief effect, and the Max/Min/Median/Mode filters for noise reduction or morphological cleanup on masks and rough images.

Installing it

ComfyUI Manager: search "Comfyui-ergouzi-Nodes", install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/11dogzi/Comfyui-ergouzi-Nodes.git

Pillow ships as a standard ComfyUI dependency already, so there's nothing extra to install for this node specifically - no models, no separate pip packages. Same note as the rest of the pack: the English repo is the frozen branch, with active development continuing in a separate Chinese-first sibling repo.

Common issues & troubleshooting

Effect looks too strong or too weak and there's no slider to fix it. That's expected - these are fixed Pillow presets, not adjustable filters. If you need finer control than a preset gives you, chain the same filter twice for a stronger effect, or reach for a different node in your setup that exposes an actual strength parameter (a proper Gaussian blur node with a radius input, for instance).

Output looks identical to the input. Some of these filters (ModeFilter, FIND_EDGES on a low-detail image) produce genuinely subtle results depending on the source image - try a filter with an obviously strong signature, like EMBOSS, to confirm the node is actually applying something before assuming it's broken.