How to do it, step by step
- Work out what lit the photo Warm bedside lamp, kitchen downlights, a candle, a sunset through a window, or a mix. Mixed sources are the hardest case, because one correction cannot be right for two temperatures at once.
- Find a neutral object in the frame A white shirt, a plain wall, a sheet of paper, the white of an eye. This is the reference the correction hangs on. If nothing in frame is neutral, you are guessing, so correct by eye against a face you know.
- Set white balance from that neutral In Lightroom or Snapseed, use the white balance eyedropper on it. In Apple Photos, drag Warmth down until the neutral stops looking beige. Move in small steps and stop as soon as the reference is neutral.
- Fix the tint axis as well as the temperature Temperature moves blue against yellow; tint moves green against magenta. Fluorescent and cheap LED bulbs add green, and skin corrected on the blue-yellow axis alone still looks unwell. Add magenta until the face looks like a face.
- Correct skin separately with HSL Open the colour mixer, take the orange or red channel, shift its hue a few points toward yellow, drop its saturation by five to ten, and raise its luminance slightly. This targets skin without touching the wooden table or a warm jumper.
- Compare against something you know Look at a familiar object in the frame, or an outdoor photo of the same person. Screens drift warm at night, so a correction made at midnight often looks too blue in the morning.
Colour temperature, in kelvin, and what each source does to skin
Every light source has a colour, measured in kelvin. Low numbers are orange and high numbers blue, the opposite of how we talk about warm and cool. Camera colour science is built around daylight at 5500 K, and every source further from that pushes the image further from neutral.
| Light source | Colour temperature | What it does to skin |
|---|---|---|
| Candle flame | about 1900 K | Deep orange, hues collapse, no colour separation in the face |
| Dimmed bulb, sunset indoors | 2200 K | Strong orange cast, red shadows |
| Tungsten and incandescent | 2700 K | The classic orange-skin culprit, and the default warm LED colour |
| Halogen, warm restaurant LED | 3000 K | Same problem, slightly cleaner greens |
| Cool white LED, office fluorescent | 4000 K | Less orange, but adds green on the tint axis, so skin looks sallow |
| Direct daylight | 5500 K | Neutral, what camera profiles are designed around |
| Overcast sky | 6500 K | Slightly cool and flat, easily corrected |
| Open shade, blue hour | 7500 K and up | Blue cast, skin drains to grey |
Dimming makes it worse: it lowers the filament temperature as well as the brightness, so a bulb at 30 per cent sits near 2200 K. That is why a nice restaurant photographs more orange than it felt.
What white balance actually does, and where it goes wrong
White balance is the camera guessing the colour of the light and applying the opposite shift. Under a 2700 K bulb everything arrives too orange, so the camera adds blue back to make white things white. Guess 4000 K when the light was really 2700 K and it adds only part of the blue it needed to. What remains is an orange cast.
Automatic white balance guesses by assuming the whole scene, averaged together, is neutral grey. That assumption fails constantly:
- Warm rooms. Wooden floors, beige walls, terracotta, skin. The average really is orange, so the camera decides the light is fine.
- One dominant colour. A red jumper filling the frame, a green lawn, a blue pool. The camera corrects a cast that is not there.
- Mixed sources. A 2700 K lamp on one side and 6500 K window light on the other. There is no single right answer, so half the frame is always wrong.
Some of it is deliberate. Phone makers do not correct warm light all the way, because a fully neutralised living room looks clinical and dead, and low light processing leans warmer still. Part of the orange is a choice made on your behalf, not an error.
Why it lands hardest on skin
Skin colour comes from two pigments: melanin, running yellow-brown to deep brown, and haemoglobin, which is red. Every skin tone sits in a narrow band of hues on the warm side of the colour wheel, and the difference between tones is mostly lightness and saturation, not hue.
That narrow band is exactly where an orange cast pushes. A wall going slightly warm still reads as a wall; a face going slightly warm crosses into a hue we read as wrong: sunburnt, jaundiced, fake tan. And because the whole face shifts together, the variation that makes skin look like skin, redder cheeks against a yellower forehead, flattens into one orange.
The effect is not equal across skin tones. Deeper tones already sit further into the orange-red region, so a cast has less room before it reads as unnatural and it flattens the shading on the face. Very pale skin shows it first in the shadows, which go ruddy before the highlights do.
Why correcting the whole frame ruins the photo
The obvious fix is to drag the temperature slider left until the face looks right. That slider is a global operation: it applies a gain to the red and blue channels across every pixel in the image. Take 1200 K of warmth out to rescue a face and you also take it out of everything else.
- The outfit changes colour. A camel coat goes grey-green, gold jewellery goes silver, a burgundy dress goes plum.
- The room stops looking lit. The pool of warm light around a lamp is the thing that makes a photo feel like evening. Neutralise it and the lamp looks switched off.
- Wood, brick and candles die. Every warm material in the frame was warm in real life, not just in the photo.
The correct move is selective: correct the hue on the skin and leave the rest of the frame at the temperature it was lit at. A warm room with correctly rendered faces is what your eye saw at the time, because your visual system does this adaptation automatically. That gap is why the photo looks more orange than the evening felt.
ShutterAI's True to You works on exactly this axis: it corrects colour casts on skin specifically, so every skin tone in the frame reads correctly while the lamp, the wood and the outfit keep the colour they had.
Fixing a photo you have already taken
Order matters. Get the global temperature roughly right first, then treat skin, then check the tint.
- Apple Photos. Edit, then Warmth and Tint. Blunt but quick. With no skin-only control, under-correct: take out about two thirds of the cast and stop.
- Lightroom mobile. The eyedropper gets you close, then Colour Mix, orange channel: hue a few points toward yellow, saturation down five to ten, luminance up slightly. The most useful skin-tone control on a phone.
- Shot in ProRAW? White balance is an editable parameter rather than a baked-in one, so a raw file re-lights from 2700 K to 5500 K with no penalty. On a HEIF, the correction stretches data that has already been discarded, so heavy moves leave banding in the shadows.
One warning about tint. Fluorescent and cheap LED light has green in it, and correcting on the blue-yellow axis alone leaves faces looking ill rather than orange. People then pull the temperature slider further, which makes it worse. Add magenta instead.
Stopping it at the source
Prevention beats correction, and it is free.
- Use one type of light. Mixed temperatures cannot be fixed globally. Turn the lamp off and use the window, or close the curtains and use the lamp.
- Shoot within two metres of a window in daylight. Window light is roughly 6000 to 7000 K, soft, and the easiest light in the world to get right.
- Change the bulbs. Swapping 2700 K bulbs for 4000 K in the room you shoot in most costs a few pounds and removes the problem.
- Turn Night Shift off before shooting. It does not affect the file, but it warms the screen, so you approve photos that are more orange than you think.
- Use ProRAW when the light is bad. Not for everything, the files are enormous, but for a candlelit dinner it is the difference between a fixable photo and an orange one.
Questions people ask
Why do my photos look orange?
Because indoor bulbs burn around 2700 kelvin while cameras are built around 5500 K daylight, and automatic white balance only partly corrects the difference. The residual warmth lands on skin, which is already in the warm part of the colour wheel, so faces read orange.
How do I fix orange skin tone without ruining the rest of the photo?
Correct only part of the cast globally, then target skin with the orange channel in a colour mixer: hue slightly toward yellow, saturation down five to ten points, luminance up a touch. A global temperature drag alone drains warm clothing, wood and lamplight.
Does ProRAW help with colour casts?
Yes, more than any other setting. White balance in a raw file is an editable parameter rather than baked into the pixels, so you can re-set a 2700 K scene to neutral with no banding. The cost is roughly 75 MB per photo.
Why do restaurant photos look worse than the restaurant did?
Dimmed incandescent bulbs drop to around 2200 K, and your eyes adapt to that while the camera only partly does. You saw a warm room with normal faces; the sensor recorded orange faces in a warm room.
Is the orange cast worse on some skin tones?
It affects all skin tones, but differently. Deeper tones sit closer to orange to begin with, so a cast has less room before it looks wrong and it flattens the shading on the face. Very pale skin usually shows it first in the shadows, which go ruddy.