Get Tags Above Threshold

This is the plain, scoped version of LogicUtils' tagger nodes - one step down from GetAllTagsAboveThresholdNode, which is deliberately built to pull in everything the model tags an image with. Going by the name, this one is meant for the more everyday case: the descriptive tags - pose, clothing, expression, setting - without necessarily pulling in every category the model can produce. If you want character names specifically, GetCharactersAboveThresholdNode covers that; if you want everything but characters, GetAllTagsAboveThresholdNode's "except character" sibling covers that. This one sits in between, closer to what most people mean when they say "just tag this image."

The practical use is captioning. Danbooru-vocabulary models - Illustrious, NoobAI, Pony - were trained on exactly this comma-separated tag format, so a WD-style tagger node like this is the standard way to caption a LoRA training set for them, instead of writing prose captions that those models never learned to read.

How it works

model_name (default EVA02_Large) picks the classifier backbone, sharing the same option set as every other tagger node in the pack: ViT_Large, SwinV2, ConvNext/ConvNextV2, MOAT, ViT, and _v3 revisions of a few of those. threshold (default 0.4) sets the confidence bar a tag needs to clear. replace, following the convention this whole family shares, most likely governs whether underscores in raw tags become spaces in the output - a small but common toggle in WD-tagger tooling generally.

Inputs and outputs

• image - the IMAGE to tag.
• threshold (default 0.4) - confidence cutoff for inclusion.
• model_name (default EVA02_Large) - the tagger backbone.
• replace (default false) - likely the underscore-to-space toggle.

Output is a single STRING - a comma-separated tag list.

Installing ComfyUI-LogicUtils

ComfyUI Manager: Install Custom Nodes → search "ComfyUI-LogicUtils" → install → restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/aria1th/ComfyUI-LogicUtils

Restart ComfyUI. This is one of the pack's real ML nodes, unlike its pure-logic siblings - expect a weights download the first time you use a given model_name, and check LogicUtils' auto-install hook (COMFYUI_LOGICUTILS_AUTO_INSTALL=1) or requirements.txt if it errors on first run with a missing package.

Worth knowing

The README documents none of the semantic differences between this node and GetAllTagsAboveThresholdNode - the split is inferred from naming, not confirmed against source, so if your output looks identical to the "all tags" version on a given image, that's plausible rather than a bug. This is aria1th's (AngelBottomless's) LogicUtils pack, from the same person who trains Illustrious XL - almost certainly why a full Danbooru tagger family lives inside what's otherwise an unrelated grab-bag of logic nodes. Public commits stopped in January 2026, so don't expect this ambiguity to get cleared up in a future release.