Verbing Styler

VerbingStyler adds an action to your prompt. Its dropdown is 3,501 gerunds - -ing verbs like abandoning, abating, abiding, abolishing - and picking one folds that action into your text. Like NounsStyler and PrepositionStyler, it's a grammar styler: a word picker built for composing phrases, not an aesthetic preset. A verb is what makes a subject do something, so this is the node that turns "a dancer" into "a dancer leaping."

Of the grammar trio, verbs are the most immediately useful on their own, because an action word actually changes the pose and dynamics of a subject in a way a bare noun or preposition doesn't. Still, the real payoff is stacking it with the noun and preposition stylers to build a full relational phrase, using the combo nodes' offset feature to place each word where you want it. Solo, it's a quick way to inject motion; combined, it's a sentence machine.

How it works

The usual MileHighStyler substitution: the chosen verb goes into a template with a {prompt} placeholder, your prompt drops in, and the added action word gets automatically weighted up (around 2.0) so the motion pushes while your subject stays put. No inference, just a templated insert.

The inputs that matter

• text_positive - your prompt.
• text_negative - your negatives.
• verbing - the 3,500-item dropdown; none contributes nothing.
• log_prompt - on by default, prints the finished prompt to the console.

Outputs are text_positive and text_negative, feeding a CLIP Text Encode pair.

Installing it

ComfyUI Manager, search ComfyUI_MileHighStyler. If Manager doesn't surface it (it's an older, niche pack), clone it:

cd ComfyUI/custom_nodes
git clone https://github.com/TripleHeadedMonkey/ComfyUI_MileHighStyler

Restart ComfyUI. No models, no dependencies - JSON templates, lightweight even with the big list.

Common snags

The 3,500-entry dropdown is easier to live with than the noun list's 90k, but still, type to filter rather than scroll. The other thing: a strong action verb can fight your composition - "leaping," "falling," "spinning" all imply motion blur and dynamic framing the rest of your prompt might not want. Keep log_prompt on and sanity-check what got assembled.

Worth remembering that not every gerund in a 3,500-word list makes visual sense - plenty of them (abating, abrogating) describe abstractions the model can't really depict. Stick to actions with a clear physical pose behind them and you'll get far more out of it.

Pack-wide reality: this is CLIP-era tech, best on the SDXL lineage. On a 2026 LLM-encoded model (Flux 2, Z-Image, Anima) the weighting is discarded, and those models handle actions and word order natively - "a dancer leaping mid-air" in plain prose beats a weighted verb token. Use VerbingStyler on SDXL and its finetunes, where the mechanism actually does something.