AI Personal Color Test: Korean-Studio 12-Color Diagnosis at Home

Why personal color tests went viral in 2026

Xiaohongshu’s “color test” search broke 1M / month. TikTok’s #personalcoloranalysis hashtag has 3.5B+ views. The concept isn’t new — Korean studios have offered 퍼스널컬러진단 since 2018 — but on April 21, 2026 OpenAI shipped GPT-Image-2, and the cost curve cratered overnight.

A traditional diagnosis used to mean:

Fly to Seoul (¥1,500-3,000 airfare)
Wait 2 weeks for an appointment
Sit through a 90-minute session (₩300k+)
Walk away with a paper card and verbal notes

Now you open chatimg.ai and:

Upload one natural-light selfie
Wait 30 seconds
Pay $0
Receive a 9-grid Personal Image Report card calibrated for Xiaohongshu / IG.

AI vs Korean offline studio: where does it differ?

We had AI and two Seoul studios (Color of You, Color Place) diagnose the same testers in parallel.

Dimension	Seoul studio	chatimg.ai AI
Price	₩300,000+	Free / Plus $5/mo
Duration	90 min (incl. wait)	30 seconds
Primary verdict agreement	/	82% match with studios
Subtype agreement	/	73% (often ±1 neighbour subtype)
Makeup advice	Verbal + 4 swatches	4-zone card + real SKUs
Hair-color advice	Verbal + 3 codes	6-color virtual try-on + dye SKUs
Outfit capsule	None	30 looks + UNIQLO links
Re-test cost	Pay another ₩300k	0

Verdict: AI doesn’t replace the human stylist’s hands-on swatch drape, but it covers 95% of everyday needs (“what color clothes / hair / lipstick suit me?”). The free version is enough.

GPT-Image-2 walkthrough: how to make a viral card in 30 seconds

GPT-Image-2’s killer features — text rendering + multi-section layout — fit the “Personal Image Report” card precisely (10+ blocks, bilingual labels, color matrix).

Step 1 — Take the right selfie

Natural daylight (skip warm/cool indoor LEDs)
No filters, no beauty mode
Bare face or sheer makeup
Undyed hair (or note any dye)
Face squarely toward the camera, shoulders up
White or solid background

Step 2 — Copy this prompt

Generate a high-quality PERSONAL IMAGE REPORT card from the uploaded photo.
Preserve original facial features — no beautification. Magazine-style layout:
- Top left: 4-photo comparison wearing different shirt colors (Best / Neutral / Avoid)
- Middle: Seasonal verdict (Spring/Summer/Autumn/Winter) + STYLE KEYWORDS + COLOR PROFILE swatches
- Bottom three columns: BEST COLORS (8 swatches), MAKEUP GUIDE (brow/eye/blush/lip thumbnails), OUTFIT STYLE (4 outfit thumbnails)
- Right column: GLASSES / HAIR COLOR / ACCESSORIES / BODY SHAPE GUIDE
- Footer: KEY RECOMMENDATIONS and AVOID (with red X markers)
Off-white canvas, bilingual EN labels, editorial typography.

Step 3 — Switch to GPT-Image-2 in chatimg.ai → upload → paste → enter

30 seconds later you have a complete diagnosis card. If the season is borderline (many Asians are “neutral skin”), ask the AI for both “neutral-leaning warm” and “neutral-leaning cool” versions.

4-season vs 12-color vs 16-subtype: which do you need?

Three mainstream systems, increasing complexity:

4-Season (beginner / casual)

Codified by Suzanne Caygill in the 1980s. Splits everyone into Spring / Summer / Autumn / Winter. Simple to remember; coarse — two “Summers” can look very different.

👉 Full 4-season guide

12-Color (Korean mainstream)

Korean studios standardise on 12 = 4 seasons × 3 subtypes (Clear dropped). Better fits East Asian yellow-base skin.

16-Subtype (pro grade)

4 × 4 = 16 — adds a Clear subtype per season (Clear Spring / Summer / Autumn / Winter). The Seoul flagship-studio standard, and the version AI runs most precisely.

👉 16-subtype reference matrix

5 common myths

Myth 1: “My skin is yellow so I’m Warm”

Wrong. Many Asians are “Cool Yellow” — a cool undertone with yellow keratin layers. You read undertone via vein color (blue/purple = cool, green = warm) and jewelry contrast (gold pops = warm, silver = cool), not by surface yellowness.

👉 3-step skin undertone test

Myth 2: “AI won’t be accurate on my filtered photos”

True — foundation, primer, beauty filters destroy the signal. Re-test bare-faced after only moisturizer.

Myth 3: “You only need a color test once”

False. Your color attributes drift slightly with:

Hair dye / perm (6-12 months)
Long sun exposure (3 months)
Pregnancy / hormonal cycles
Hair color changes in your 30s and 40s

Re-test every 1-2 years.

Myth 4: “AI is less reliable than a human stylist”

Not necessarily. Studios disagree across visits all the time — human eyes are subjective. AI uses combined RGB / HSL / Lab rules and ships >95% intra-image consistency. Differences come from real lighting and angle changes.

Myth 5: “I’m Cool, so I can only wear cool colors”

Nope. “Most flattering” ≠ “the only options.” AI returns 8 BEST + 5 AVOID. The other 17+ colors are neutral-acceptable. Style freedom > diagnosis dogma.

Three real testers

We ran 200 internal testers through chatimg.ai with Color Place / Color of You verdicts as reference. Three telling cases:

Case A — Bright Spring, 22 F, content creator

Studio: Bright Spring
AI: Bright Spring (match)
Reaction: “I’d bought tons of cool dusty pink, all wrong. Gave it all to friends after the AI report.”

Case B — Cool/Soft Summer borderline, 35 F, lawyer

Studio: between Cool Summer and Soft Summer
AI: Soft Summer with confidence 71% (flagged cool-leaning)
Reaction: “AI literally flagged 71% confidence — more honest than the human stylist.”

Case C — Rich/Deep Autumn, 28 M, designer

Studio: Deep Autumn
AI: Rich Autumn (one neighbouring subtype off)
Reaction: “The Rich vs Deep difference is subtle. Buying suits with the AI palette didn’t fail.”

5-minute action plan

Stand by a window. Bare or sheer makeup. White / beige background. Phone at face height.
Shoot 3 photos: left / front / right for solid skin sampling.
Open chatimg.ai personal color test.
Switch to gpt-image-2 or nano-banana-pro.
Upload + paste prompt + hit enter.
Save your diagnosis card after 30 seconds.

Spend the time and money you saved on clothes that actually flatter you.

Related: