Generates a randomised, structured physical description of a person for use in image generation prompts. Every attribute is drawn from curated JSON data files tuned for realistic, unambiguous output with SDXL and Flux models.

Purpose

Diffusion models drift toward a narrow set of "default" faces when a prompt leaves appearance vague, so batches of people start to look the same. This node feeds the model a specific, well-formed human description on every run: nationality, age, complexion and skin texture, eye colour, face shape and distinctive features, hair, facial hair, build, expression, accessories, and makeup. Pinning down those traits, and varying them per seed, pushes the model away from its defaults and toward more unique, individual people across a batch. Lock the traits you care about and let the rest randomise to explore variety without losing control.

The node will be located under Add Node > utils. Node name: Random Person Description.

Example output (male):

74 year old Pakistani male, oval face, light complexion, wrinkled skin, pale light grey eyes,
aquiline nose, wide-set lips, a scar through one eyebrow, ash grey shaved hair worn taper fade,
a van dyke beard, athletic build, glasses

Example output (female):

43 year old Russian female, angular face, light medium complexion, smooth skin, natural grey eyes,
button nose, wide-set lips, a faint scar through one eyebrow, red very short hair worn twist out,
muscular build, reading glasses, bold lipstick

Installation

ComfyUI Manager (recommended)

In ComfyUI, open Manager > Custom Nodes Manager, search for ComfyUI_RandomPerson, click Install, then restart ComfyUI.

Manual install

Copy the ComfyUI_RandomPerson folder into your ComfyUI/custom_nodes/ directory
Restart ComfyUI

The node appears in the Add Node menu under utils > Random Person Description.

Output Pins

Pin	Contents	Example
`description`	Full comma-separated description string	`43 year old Russian female, angular face, ...`
`sex`	Resolved sex	`female`
`age`	Age as a plain number	`43`
`nationality`	Nationality descriptor	`Russian`
`complexion`	Skin tone and texture	`light medium, smooth skin`
`face`	Combined face attributes (shape, eyes, nose, mouth, distinctive feature)	`angular face, natural grey eyes, button nose, wide-set lips, a faint scar through one eyebrow`
`hair`	Combined hair description	`red very short hair worn twist out`
`facial_hair`	Beard / moustache style (empty for clean shaven)	`a van dyke beard`
`body_type`	Build	`muscular`
`expression`	Facial expression (empty for neutral)	`a soft smile`
`accessories`	Eyewear / jewellery (empty for none)	`reading glasses`
`makeup`	Makeup style (empty for none / male)	`bold lipstick`
`seed`	The seed used (useful for reproducibility)	`481046155`

Wire description directly into a text prompt node, or use individual pins to route specific attributes to other parts of your workflow.

Controls

Seed & Randomise

Control	Purpose
`seed`	Fixed seed, use the same value to reproduce an identical result
`randomize`	When on, ignores the seed and picks a new random person every run

Sex

Dropdown: random / male / female

When set to male or female, the dropdown options for hair style, hair colour, face shape, nose, mouth, body type, facial feature, facial hair, and makeup automatically filter to show only sex-appropriate options. Facial hair and makeup are effectively sex-specific: females never grow facial hair and males have no makeup options.

Per-Category Controls

Every attribute category (nationality, complexion, skin texture, eyes, face shape, nose shape, mouth shape, facial feature, hair colour, hair style, hair length, facial hair, body type, expression, accessories, makeup) has three controls:

Control	What it does
mode	How the value is chosen, see Mode Options below
allow_list	Comma-separated list of values to draw from (used when mode = `allow_list`)
fixed	A specific value to always use (used when mode = `fixed`)

Mode Options

Mode	Behaviour
`random`	Pick any value from the full list for the resolved sex
`allow_list`	Pick randomly from only the values you specify in the allow_list field
`fixed`	Always use the value selected in the fixed dropdown
`off`	Skip this attribute entirely, it will not appear in the description

Tip: turning off attributes: Set any category's mode to off to remove it from the output completely. For example, if you don't want a body type in the description, set body_type_mode to off.

Default modes

Core identity categories (nationality, complexion, eyes, face shape, nose shape, mouth shape, hair colour/style/length, body type, plus age and sex) default to random. The optional "flair" categories default to off so the base person stays clean and realistic, and you opt them in per category:

skin_texture, face_feature, facial_hair, expression, accessories, makeup

The example outputs above have several of these enabled to show the full range.

Allow List: Custom Values

The allow_list field accepts comma-separated values. These can be:

Labels from the JSON data, e.g. auburn, golden blonde, platinum blonde, the node will pick randomly from the matched entries
Custom values not in any list, e.g. fire engine red, pastel pink, unknown tokens are treated as literal descriptors and added to the pool

This means you can mix standard and custom values freely:

auburn, copper red, fire engine red

Age

Control	Purpose
`age_mode`	`random`, `fixed`, or `off`
`age_min`	Minimum age (0 = no lower bound). Cannot go below 21.
`age_max`	Maximum age (0 = no upper bound, defaults to 90)
`age_fixed`	Exact age to use when mode = `fixed`. Values below 21 are clamped to 21.

Extra Attributes

A free-text area at the bottom of the node. Enter any custom descriptors separated by commas, they are appended to the end of the description exactly as written.

wearing glasses, tattoo on left arm, silver hoop earrings

This is the right place for clothing, accessories, expressions, or any detail not covered by the built-in categories.

Available Values

Nationality (40)

American, Algerian, Argentinian, Australian, Bangladeshi, Brazilian, British, Canadian, Chinese, Colombian, Dutch, Egyptian, Ethiopian, Filipino, French, German, Ghanaian, Greek, Indian, Iranian, Italian, Jamaican, Japanese, Kenyan, Korean, Mexican, Moroccan, Nigerian, Pakistani, Peruvian, Polish, Russian, South African, Spanish, Sudanese, Swedish, Thai, Turkish, Ugandan, Vietnamese

Complexion (15)

porcelain, fair, light, light medium, warm beige, warm olive, olive, medium, warm medium, tan, golden brown, medium brown, deep brown, deep, ebony

Skin Texture (10)

smooth, clear, dewy, glowing, freckled, fine lines, wrinkled, sun-spotted, acne scars, ruddy

Skin texture is separate from complexion (tone). It appends after the complexion in both the description and the complexion pin, e.g. tan, smooth skin.

Eye Colour (23)

brown, dark brown, light brown, near black, hazel, hazel green, hazel brown, amber, blue, light blue, dark blue, ice blue, grey blue, steel blue, green, light green, dark green, olive green, blue green, grey, pale grey, dark grey, grey green

Eye colour descriptions are deliberately grounded: e.g. green outputs as muted natural green eyes and amber as warm amber-brown eyes, to prevent image models rendering oversaturated or unrealistic eye colours.

Hair Colour

Male (25): jet black, black, dark brown, chestnut brown, warm brown, brown, medium brown, light brown, auburn, copper red, red, dark red, warm honey, golden blonde, blonde, dark blonde, light blonde, sandy, ash blonde, salt and pepper, silver, ash grey, dark grey, grey, white

Female (27): same as male plus strawberry blonde, platinum blonde

Hair Style

Male (31): afro, bowl cut, braids short, bun, buzz cut, caesar cut, coiled natural, cornrows, crew cut, cropped short, curly, curly medium, dreadlocks, faux hawk, french crop, locs long, locs short, low ponytail, messy medium, mohawk, mullet, pompadour, quiff, side part, skin fade, slicked back, straight, taper fade, textured crop, undercut, wavy

Female (39): afro, blunt cut, braid over shoulder, braids, bun, chignon, chin length bob, coiled natural, cornrows, cropped short, curly, curtain bangs, dreadlocks, faux hawk, fishtail braid, french braid, half up, high ponytail, jaw length bob, layered, locs, loose curls, loose waves, low ponytail, messy, mohawk, mullet, pixie, pulled back loose, shag, space buns, straight, textured crop, tight curls, twist out, updo, wavy, wispy, wolf cut

Hair Length

Male (7): shaved, very short, short, ear length, medium, shoulder, long

Female (9): shaved, very short, short, ear length, chin length, medium, shoulder, long, very long

Face Shape

Male (8): oval, square, rectangular, round, oblong, diamond, triangular, angular

Female (8): oval, round, square, heart, diamond, oblong, triangular, angular

Nose Shape (14 each)

Male: straight nose, aquiline nose, broad nose, rounded nose tip, curved nose bridge, snub nose, roman nose, wide nose, flat nose, narrow nose, upturned nose, high nose bridge, low nose bridge, crooked nose

Female: straight nose, upturned nose, button nose, broad nose, aquiline nose, snub nose, narrow nose, wide nose, crooked nose, flat nose, roman nose, rounded nose tip, high nose bridge, low nose bridge

Mouth Shape

Male (9): thin lips, full lips, wide mouth, narrow mouth, downturned mouth, upturned mouth, defined cupids bow, prominent lower lip, wide set lips

Female (12): same as male plus bow shaped lips, heart shaped lips, rosebud lips

Facial Feature (distinctive marks)

Male (13): heavy brow, scar through eyebrow, deep set eyes, high cheekbones, cleft chin, strong jaw, laugh lines, gap toothed smile, dimples, freckles, mole on cheek, birthmark on jaw, weathered skin

Female (12): high cheekbones, scar through eyebrow, deep set eyes, wide eyes, cleft chin, dimples, gap toothed smile, freckles, beauty mark, birthmark, laugh lines, arched brows

Folded into the face pin alongside shape, eyes, nose, and mouth.

Body Type

Male (18): slim, lean, wiry, slight, average, medium build, athletic, muscular, stocky, broad shouldered, tall lean, tall athletic, lanky, heavyset, barrel chested, compact, softly built, dad bod

Female (19): slim, lean, wiry, petite, slight, average, medium build, athletic, muscular, tall lean, tall athletic, lanky, pear, hourglass, curvy, full figured, softly built, heavyset, stocky

Facial Hair (male)

clean shaven, light stubble, heavy stubble, short beard, full beard, goatee, moustache, circle beard, van dyke, mutton chops, soul patch

Females have only clean shaven, so a random female never grows facial hair. clean shaven outputs nothing.

Expression (11)

neutral, soft smile, warm smile, slight smile, serious, stern, pensive, confident, relaxed, intense gaze, laughing

neutral outputs nothing, so many people carry no expression text.

Accessories / Eyewear (12)

none, glasses, thin framed glasses, round glasses, thick framed glasses, reading glasses, sunglasses, stud earrings, hoop earrings, nose ring, nose stud, small ear piercings

none outputs nothing.

Makeup (female)

none, natural makeup, soft makeup, minimal makeup, bold makeup, glam makeup, smokey eye, winged eyeliner, red lipstick, bold lipstick

Males have only none. none outputs nothing.

Tips

Reproducing a result: when you find a description you like, set randomize to off and note the seed value from the seed output pin. Set seed to that value and re-queue to get the same result.

Locking some attributes, randomising others: set the attributes you want fixed to fixed mode, leave the rest on random. For example, fix nationality to Japanese and complexion to warm medium while letting everything else randomise.

Building a character type: use allow_list mode with a curated set of values per category to constrain randomness to a character archetype. For example, for a rugged older man: age_min = 45, body_type_allow_list = athletic, broad shouldered, muscular, hair_color_allow_list = salt and pepper, silver, grey.

Adding detail the node doesn't cover: use the extra_attributes text area for anything else: clothing, accessories, expression, pose hints, scars, tattoos, jewellery, makeup, etc. Each comma-separated entry is appended as its own descriptor.

Turning off attributes you don't need: if your workflow already handles body type from another node, or you don't want age in the prompt, set that category's mode to off. The attribute disappears from the description entirely.

Data Files

All values live in data/ as plain JSON files. You can edit them directly to add, remove, or change any value, no code changes needed. New entries take effect on the next ComfyUI restart.

data/
├── shared/          # used by both male and female
│   ├── nationality.json
│   ├── complexion.json
│   ├── skin_texture.json
│   ├── eyes.json
│   ├── expression.json
│   ├── accessories.json
│   └── age.json
├── male/
│   ├── hair_color.json
│   ├── hair_style.json
│   ├── hair_length.json
│   ├── facial_hair.json
│   ├── face_shape.json
│   ├── nose_shape.json
│   ├── mouth_shape.json
│   ├── face_feature.json
│   ├── body_type.json
│   └── makeup.json
└── female/
    └── (same files as male/)

Each JSON entry follows this structure:

{ "label": "auburn", "description": "auburn" }

The label is what appears in dropdowns. The description is what goes into the prompt. They can differ: for example, eye colours use grounded descriptions ("label": "green", "description": "muted natural green eyes"), and clean shaven / neutral / none use an empty description so they add nothing to the output.

A leftover face_feature.json also exists under shared/ from earlier data; it is not loaded (facial features are sex-gated). Safe to ignore or delete.

Testing

Unit tests cover the selection logic with no ComfyUI install required (they read only the bundled data/ JSON):

cd ComfyUI_RandomPerson
python -m unittest test_random_person -v

Requirements

ComfyUI (any recent version, Nodes 2.0 compatible)
Python 3.10+
No additional pip packages required

The node registers on both the legacy V1 dict API (NODE_CLASS_MAPPINGS) and the V3 schema API (comfy_entrypoint), so it loads on older and newer ComfyUI builds alike. The shared selection core is wrapped by thin adapters for each API.

Support

If you find this useful, please consider starring the repo. Stars help other people discover these nodes.