🐳Clothing Selector (Multi-Select)

The painful way to prompt clothing is to recall the exact English tag a model was trained on - is it "sheath dress" or "shift dress"? "wrap blouse" or "kimono top"? The 🐳Clothing Selector (Multi-Select) node replaces that memory game with dropdowns. It ships a curated catalog of clothing tags organized by category - dresses, tops, bottoms, swimwear, sportswear, underwear, outerwear, special - and lets you pick up to three from each via simple dropdown menus. No typos, no hallucinated garment names, no tag soup.

This is the "manual but structured" member of the Pond clothing family. Where ClothingSelectorSimple makes you type numbers and ClothingSelectorBatch rolls the dice, this one is the deliberate option: you see the full list of ~25 garments per category and hand-pick the ones you want. Each select_<category>_<n> dropdown defaults to 无 (none), so the node is inert until you actually choose something.

The inputs that matter

• select_连衣裙_1..3, select_上衣_1..3, select_下装_1..3 - the pickers, three slots per category. This is the whole node, honestly: you're choosing dress/top/bottom combinations from curated lists.
• separator - comma, space, or newline, default comma. Controls how the tags join in the output. Comma is what CLIP text encoders expect.

The category list also covers 泳装 (swimwear), 运动装 (sportswear), 内衣 (underwear), 外套 (outerwear), and 特殊服装 (special), so a full look - including under and over layers - is buildable from one node.

Outputs

Three STRING outputs, which is the part people miss at first:

• english_tags - the English-only tag string (what the model was actually trained on).
• chinese_tags - the Chinese-only version.
• combined_tags - both, joined.

Wire english_tags into your positive prompt for English-token models (SD1.5/SDXL/Flux-class checkpoints). If you're prompting a Chinese-trained model like a Qwen-based one, chinese_tags is the one that lands. The combined output is for when you want both tokens in the prompt, which some bilingual models genuinely benefit from.

Install and the honest take

From comfy_Pond_Nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/Pondowner857/comfy_Pond_Nodes
cd comfy_Pond_Nodes && pip install -r requirements.txt

Pure data, no models, instant. Honest caveats: like the whole clothing family in this pack, the catalog is heavily women's-fashion - 25 dress options versus a thinner menswear tail. And the tooltips are the author's own (they're Chinese, so this article is translating them) - when in doubt, trust that the picker list is the source of truth for what you can choose. If you want the exact same vocabulary but prefer typing numbers (or want to feed a list from elsewhere), the Simple sibling is the same data in a different skin.