AI-generated portrait of WOSHI in a black silk cheongsam in a rainy Hong Kong night market, 1997.
AI fashion model · built locally

WOSHI

Ten looks. One face. Zero API budget.

An AI fashion persona built on a Mac with open-source models — with the prompts, the rejects, and the C-dramas that inspire the looks.

The project

She doesn't exist offline.

WOSHI is generated entirely on a local machine: Qwen-Image 2.1 for the stills, LTX-2.3 for motion. No cloud render, no stock photography, no API budget. The face is locked with three reference images; everything else — city, era, light, outfit — is a prompt.

The account is a build-in-public log: one look at a time, with the model, the steps, the time, the rejects. The taste is the expensive part.

Lookbook · 001–010

Ten looks, one face.

Hong Kong rain, 1930s Shanghai, a Chengdu teahouse at six in the morning. The same woman, ten different decades.

WOSHI in a black cheongsam in a rainy Hong Kong night market, 1997.
001 · Hong Kong1997
WOSHI in an emerald qipao in a 1930s Shanghai art deco hotel lobby.
002 · Shanghai1937
WOSHI in a cream linen shirt in a Chengdu teahouse at dawn.
003 · Chengdu6am
WOSHI in a dark coat on the last Tokyo train.
004 · Tokyolast train
WOSHI in a camel coat in a Beijing hutong in winter.
005 · Beijingwinter
WOSHI in an ivory dress on a wooden boat in the Guilin mist.
006 · Guilinmist
WOSHI in a red silk dress in a 1960s Macau casino.
007 · Macau1960s
WOSHI in a fur coat on the Shanghai Bund in the 1920s.
008 · Shanghai Bund1920s
WOSHI in a Hong Kong tailor shop with a measuring tape, 1966.
009 · Hong Kongtailor, 1966
WOSHI in a grey wool coat on a quiet Seoul street, 1998.
010 · Seoul1998
How she's made

The stack.

45s

One still

Qwen-Image 2.1, 25 steps, 819×1024. Three reference images for the face. About 45 seconds per accepted image — nine of ten get rejected.

137s

One motion

LTX-2.3 turns a still into four seconds of video at 768×1024. About 137 seconds. No cloud, no render farm.

€0

One rule

No API budget. No paid models. The machine does the work; the human does the taste. Every look is generated locally.

The prompt behind Look #001
The same woman as image_1 and image_2: Caucasian, late twenties, pale skin, long copper-red hair, green eyes, sharp cheekbones. Identical face and identity. Full-body cinematic editorial fashion photograph. She walks through a rainy Hong Kong night market in 1997, wearing a black silk cheongsam, holding a small black umbrella, neon signs reflecting in puddles, looking slightly off-camera, 35mm film grain, shallow depth of field, moody teal and red palette, full body visible head to toe. No text, no watermark.
The rules

What this is — and what it isn't.

AI-generated, human-directed. She is not a real person, and the account says so.
No affiliate links, no "10x" threads, no fake brand deals.
One face, one visual world. The consistency is the product.
C-dramas, not clips. The inspiration is the aesthetic, not someone else's footage.