Art Style

If you generate anime-style images with any regularity, you have a prompt you paste into every job - masterpiece, best quality, amazing quality, ultra detailed, and friends. The Art Style node from the Prompt String Selector pack is basically a home for that block. You stash it in a text file, pick it from a dropdown, and stop retyping the same eleven tokens in every CLIP text encode.

It's part of a larger idea the whole pack is built on: treat your prompt as a set of blocks - style, character, pose, background, clothing, emotion - and keep each block in its own text file. The prompt-engineering crowd has been moving this way for a while, and for CLIP-encoded models (SD 1.5, SDXL, Illustrious, NoobAI, Pony) it maps directly onto how those models actually read: a tidy, comma-separated tag list. The Art Style selector is the top of that pyramid, since a quality/style block is the part most people share across every image.

How it works

The node reads the folder ComfyUI/custom_nodes/ComfyUI-Prompt-String-Selector/data/art_style/. Every .txt file in there becomes one entry in the dropdown; the filename minus extension is what you see. The pack ships with a single 01_Style1.txt, so out of the box you get 01_Style1 plus custom_input in the menu.

Two rules make this tick, and they apply to every selector in the pack:

• Names start with two digits and an underscore - 01_none.txt, 07_watercolor.txt. That leading number becomes the id output. Files sort by it, so it doubles as a manual ordering tool.
• Comments are legal. Anything between /* ... */ or after // in the file is stripped before the text goes to the sampler. Handy for leaving yourself a note in the file.

Inputs and outputs

Only two inputs, and only one you'll touch most days:

• art_style - the dropdown of your saved style blocks.
• art_style_custom - a multiline text box. Pick custom_input in the dropdown and whatever you type here wins, which is your escape hatch for one-off styles you don't want to commit to a file yet. (The id output is 0 in that case.)

Outputs are the standard selector pair: text (STRING, the style block, comma-terminated so it glues onto the next block) and id (INT, the file's leading number). Wire text into your CLIP text encode's positive input, or better, into one slot of the pack's Aggregator so it gets combined with your character, pose and background blocks in one place.

Install

ComfyUI Manager → search "Prompt String Selector", or:

cd ComfyUI/custom_nodes
git clone https://github.com/itom0717/ComfyUI-Prompt-String-Selector.git

Then restart ComfyUI. That's the whole install - there's no requirements.txt, no pip dependencies, nothing but the Python standard library. This is about the safest custom node install you can do.

Gotchas

The dropdown is built when ComfyUI starts, so a new .txt file doesn't appear until you restart. It's a minor annoyance but it will bite you the first time you drop a file in and wonder why the menu didn't refresh.

The README is Japanese-only, which is fine for a utility this simple, but know that the author's framing is "manage your prompts as files, switch them by ID." The Art Style node is the easiest one in the pack to get value from - it's just a better, reusable way to store your quality block.