Nouns Styler

NounsStyler is the odd one out in this pack. Instead of a tidy dropdown of ten aesthetic presets, it hands you a dictionary - 90,964 nouns, from aardvark on down. Pick one and it gets folded into your prompt. This isn't a "style" in the way LightingStyler or MoodStyler are; it's a word picker, part of the pack's grammar stylers (nouns, verbs, prepositions) meant for composing phrases rather than setting a look.

Be honest with yourself about what a ninety-thousand-item dropdown is for. You're not going to scroll it. It's useful in two situations: forcing a specific object into a scene without retyping your whole prompt, and brainstorming - flipping to a random noun to see what the model does with it. As a precision tool it's clumsy; as a "throw a wildcard object in and see" tool it has a certain charm.

How it works

Same substitution engine as the rest of MileHighStyler: the chosen noun goes into a template with a {prompt} placeholder, your text drops in, and the added word gets automatically weighted up (around 2.0). The pack's real trick for these grammar nodes is that you stack several stylers and use the combo nodes' offset feature to place words at specific spots - that's how a noun, a verb, and a preposition become a phrase. On its own, one noun is just one noun.

The inputs that matter

• text_positive - your prompt.
• text_negative - your negatives.
• nouns - the giant 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. For an SDXL split text_g / text_l setup, use NounsStylerAdvanced; this basic node gives the single pair.

Installing it

ComfyUI Manager, search ComfyUI_MileHighStyler. Manager doesn't always list this pack, so clone it if needed:

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

Restart ComfyUI. No models, no dependencies - JSON templates only, so despite the huge word list it's still lightweight.

Common snags

The 90k dropdown itself is the main friction. On some ComfyUI frontend versions a list that size is sluggish to open - type to filter rather than scroll, and expect a beat of lag. Also don't expect a lone noun to transform an image; it adds a concept, it doesn't compose a scene. That's what stacking with the other grammar stylers is for.

And the pack-wide caveat: this is CLIP-era tooling, best on the SDXL lineage. On a 2026 LLM-encoded model (Flux 2, Z-Image, Anima) the weighting on the injected word is discarded, so you'd get the same effect - more control, honestly - by just typing the noun into your prompt where you want it.