France Attire

Want a beret without the model turning it into a flat cap, or a Breton shirt that's actually striped and actually Breton? The France Attire node from ComfyUI Globetrotter Nodes writes that prompt for you. It's a small node - the smallest in the pack, honestly - but it nails the one job it has: putting the right French garment names and descriptions in front of your text encoder instead of leaving the model to guess.

What it is

One of the country nodes in manifestations/comfyui-globetrotter. Like every country node, it's generated at load time from JSON files in data/ - nothing about French clothing is written in Python. It takes a person (age, gender, hair, emotion) plus a handful of attire dropdowns and produces a single prose prompt STRING that you wire into a CLIP Text Encode (Prompt) node.

How it works

The mechanism is the pack's standard one: assemble the prompt in priority order (subject → appearance → emotion → attire → pose → setting → atmosphere → composition → quality), gender-filter the attire lists, and attach each garment's full description to its name. The one thing to know about this node: it's data-thin. No regional appearance data (region is none/random/Unspecified), no country poses, no landmarks. You get four attire categories and a few genuinely French pieces.

The inputs that matter

• Upper_Body - the best of the bunch: Breton Shirt (Marinière) and Bleu De Travail (the classic blue work jacket).
• Head - Beret, Coiffe.
• Legs - Sabots, Espadrilles. The espadrille entry alone is worth having, because most models render them as generic slip-ons.
• gender and age shape the subject line; emotion adds an expression. randomization + seed control reproducibility the same way as every node in this pack - "full" randomization ignores the seed, "off" pins it.
• lora_trigger, custom_prompt, and the experimental_llm_rewrite toggle round out the inputs; the last one needs transformers and downloads distilgpt2 on first use, so leave it off unless you want that.

Output

A single prompt STRING. Connect it to CLIP Text Encode as your positive prompt, then to KSampler as usual.

Install

cd /path/to/ComfyUI/custom_nodes
git clone https://github.com/manifestations/comfyui-globetrotter
cd comfyui-globetrotter
pip install -r requirements.txt   # optional

Restart ComfyUI, or search "Globetrotter" in ComfyUI Manager. No model downloads - the data is in the repo.

Troubleshooting

• Only a handful of attire choices: correct, and not a bug. France's data file is thin compared to India or Indonesia; if you need more French garments you can add them to data/attire/*/fr.json and restart - the pack is designed to be extended that way.
• Randomization is on by default (moderate): if outputs wander between runs, set randomization to "off" and fix the seed.
• experimental_llm_rewrite silently no-ops without transformers installed - it won't error, it just skips.

It's a small node with a small job. Reach for it when you specifically want French streetwear or traditional pieces in a character prompt and you're tired of the model improvising.