肖像大师：化妆

There's a whole genre of portrait prompt that lives or dies on one word: makeup. Too little and the model applies its own generic gloss; too much and you're in drag territory when you wanted "quietly done-up." PortraitMasterMakeupCN (肖像大师：化妆) is the smallest module in the Portrait Master v3 split, and it's the one that ends the guesswork - a style dropdown, a color dropdown, and six checkboxes, none of which require you to remember the exact phrasing for "soft glam."

It's part of comfyui-portrait-master-cn, the Chinese localization of Stefano Flore's Portrait Master. Pure prompt assembly - no models, no dependencies, nothing to download past the clone.

How it works

Like its sibling modules: a dropdown resolves to an English prompt phrase, and the node emits (makeup-style:1.05) wrapped tokens. The interesting part is the six booleans at the bottom - 眼影 (eyeshadow), 眼线 (eyeliner), 睫毛膏 (mascara), 腮红 (blush), 口红 (lipstick), 唇彩 (lip gloss). Each one independently injects (eyeshadow make-up:1.05) and the like. So instead of typing "smokey eyes, winged eyeliner, nude lip" you check three boxes and the node writes it for you, consistently, every run.

The inputs that matter

• 化妆风格 (makeup style) - 45 entries including 中式古风妆 (Chinese ancient-style), 中国戏曲妆 (Chinese opera), 埃及艳后妆 (Cleopatra), 动漫妆 (anime). This is where the Chinese localization genuinely pays off, because those are styles you'd otherwise be describing from memory in English.
• 化妆颜色 (makeup color) - 61 color options, emitted as (color make-up color:1.05).
• The six toggles above - the real workhorses. 口红 and 眼线 are the two beginners reach for first.

Optional text_in and seed work exactly as in the other modules; output is a single text_out STRING.

Installing it

cd ComfyUI/custom_nodes
git clone https://github.com/DropFan/comfyui-portrait-master-cn

Restart ComfyUI, or search comfyui-portrait-master-cn in ComfyUI Manager. No requirements.txt, no model files, no API keys.

Where people get burned

• The toggles override nothing - they add. With 启用 on, every checked box gets its own token, so checking all six produces a full beat-face that most checkpoints render as heavy glam. If you want a light look, check two, not six.
• The makeup style and color emit as separate weighted tokens, and on SDXL-lineage models the (...:1.05) wrappers pull their weight. On LLM-encoder models (Flux, Z-Image) the weights are mostly discarded - the vocabulary still lands, the emphasis doesn't.
• It only emits positive tokens. There's no "no makeup" output here; for a bare-faced look that's the SkinDetails module's 素颜 (bare face) slider doing the opposite job.

The honest take

It's a utility module, not a headline one - you'll reach for it when you're building a character and want the same base beauty treatment across a batch of generations. The booleans are the selling point: they turn "makeup" from one ambiguous token into six explicit ones the model can actually honor. Small, boring, and quietly the most reliable part of the pack.