01 · Reality check
What ChatGPT actually does well here
Good at
- Explaining what undertone, contrast and depth actually mean
- Suggesting a palette and explaining why each colour works
- Recommending specific colours to try in clothing and makeup
- Giving you the vocabulary to shop more deliberately
- Comparing two options when you describe how each looked on you
Not the right tool for
- Judging colour accurately from a photo, since cameras shift white balance
- Accounting for your screen, which may render the photo differently again
- Replacing professional draping, where fabric is tested against your face
- Certainty: it will sound confident even when the photo is misleading
- Knowing how a specific brand's fabric or shade will actually look
02 · The method
Step by step
- 1
Take the photo in daylight
Face a window, no direct sun, no artificial light mixed in. Artificial light is the single biggest source of wrong answers here.
- 2
Strip out everything that distorts
No filter, no makeup, hair back, plain neutral background, and wear white or grey. Coloured clothing reflects onto your skin and changes the reading.
- 3
Ask for reasoning, not just a season
Have it explain what it observes about undertone, contrast and depth before naming a season. The reasoning is more useful than the label, and it lets you judge whether it saw what you see.
- 4
Ask what would change its mind
Have it state its confidence and what it is uncertain about. This is the step that separates a useful hypothesis from a confident guess.
- 5
Test against real fabric
Hold actual clothing in the suggested colours up to your face in daylight. Your eye comparing real fabric beats any photo-based analysis.
- 6
Keep what works, discard the label
The point is knowing which colours make you look well, not being assigned a season. If a recommended colour looks wrong on you, it is wrong, whatever the analysis said.
03 · Use this now
Copy-paste prompt for a colour analysis
Here is a photo taken in natural daylight, no filter, no makeup, plain background. Before giving me a season, describe what you actually observe: my apparent skin undertone, the contrast between my hair, skin and eyes, and my overall depth. Explain the reasoning for each. Then suggest a palette of 12 colours with a short note on why each works, and 5 colours to avoid and why. State your confidence level and what about this photo makes you uncertain, including anything the lighting or camera may be distorting. Finally, tell me exactly how to verify this myself with real fabric in daylight.
04 · Avoid these
Common mistakes
- Using a photo taken in artificial or mixed light, which is the main cause of wrong results.
- Wearing makeup or coloured clothing, both of which change how your skin reads.
- Treating the assigned season as fact when the photo may have misled it.
- Buying a wardrobe based on the result before testing colours against your face.
- Assuming confident phrasing means accuracy. It will sound certain regardless.
05 · Questions
Frequently asked questions
Can ChatGPT do a color analysis?
It can analyse an uploaded photo and suggest your undertone, contrast level and a palette, with reasoning. Whether that analysis is correct depends heavily on your photo, because camera white balance and lighting distort skin colour. Treat it as a useful starting hypothesis to test, not a definitive result.
How accurate is ChatGPT color analysis?
Less accurate than it will sound. The fundamental problem is not the model but the photograph: lighting, white balance and screen calibration all shift how skin reads. Professional analysis uses physical fabric against your face in controlled light for exactly this reason. Verify any result with real clothing in daylight.
What photo should I upload for the best result?
Natural daylight facing a window, no direct sun, no artificial light mixed in, no filter, no makeup, hair pulled back, plain neutral background, wearing white or grey. Coloured clothing reflects onto your face and is a common reason results come back wrong.
Can ChatGPT tell me my season?
It will give you one, and the reasoning behind it is often genuinely useful. Be aware it is inferring from a photograph with all the colour-accuracy problems that involves. The practical value is the palette and the explanation, not the label itself.