Styles 📋

This is the style-and-mood dropdown of the pack. In the README's structure, "Styles" is the folder for art styles and artist styles; in the shipped node it currently surfaces a lighting styles list - a set of named lighting looks (Chiaroscuro, Cinematic Lighting, Backlighting, Bright and Sunny, and so on) you pick or randomize into your prompt.

Styles is part of WildPromptor by 1038lab (also the author of the ComfyUI-RMBG masking pack).

How it works

Same folder-reader design as every list node here. The data/Styles folder holds .txt files, each rendered as a dropdown. Out of the box that's lighting_Styles (48 entries). Each dropdown opens with:

• ❌disabled - skip it
• 🎲Random - random line, seed-reproducible
• 🔢ordered - cycle across a batch

Whatever's enabled is resolved and output.

Because the whole system is folder-driven, this node is a natural home for your own style lists - drop an artists.txt or art_styles.txt into data/Styles (or build one with Custom List Manager) and it becomes another dropdown right here. That's how you'd flesh out the "artist styles" half the README describes.

Inputs and outputs

The dropdown(s) are the interface; the extra controls are batch_size, seed, and allow_duplicates. Output is prompt, a STRING list, which you feed into the styles slot of Prompt Concat or straight into a text encoder.

A note on what actually moves the needle

Lighting terms are a good use of prompt space - "chiaroscuro," "backlighting," "city lights" are concrete, describable looks that most models handle well, and they read fine across model families. Artist-name styling is more model-dependent: it's alive and well on SDXL-lineage checkpoints (Illustrious, Pony) trained on tagged art, and hit-or-miss on LLM-encoded models. If you add an artists list here, expect it to behave very differently depending on what you're generating with.

There's also some overlap to be aware of: the pack's Virtual node carries its own lighting list, and the All-in-One node folds both a Styles - lighting_Styles and a Virtual - lighting dropdown into the same interface. If you enable lighting in two places at once you'll get two lighting phrases stacked in one prompt, which is usually more than you want. Pick one source for a given aspect and leave the other disabled - it keeps the prompt lean, and lean is what current models reward.

Installing it

ComfyUI Manager: search ComfyUI-WildPromptor, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/1038lab/ComfyUI-WildPromptor

then restart. Under 🧪AILab/🧿WildPromptor. No downloads.

Common issues

• Only one dropdown shows - that's expected if data/Styles has one file. Add more .txt files to get more style categories, then restart.
• My added style list is scrambled - custom list ordering isn't always respected in the UI; use 🔢ordered mode if sequence matters.
• Artist names do nothing - that's model-dependent; artist styling needs a checkpoint trained to recognize them, which is mostly the SDXL lineage.
• Can't see the random pick - chain Show Prompt downstream.