When the colour on screen is not the colour in the box

The chain, and where it actually breaks

It starts with the light in the room. A shop lit by a mix of window daylight, warm ceiling spots and a cool LED strip is illuminating the garment with three different colours at once, and the fabric reflects all of them. Then the camera guesses which of those is white. Then the phone applies its own processing, which is tuned to make photos look pleasant rather than accurate, and often warms skin tones and boosts saturation. Then the image is re-encoded by the platform. Then the buyer looks at it on a phone with the brightness down and a colour-shifted night mode running.

Most sellers assume the buyer's screen is the problem, and it does contribute, but it is the one link nobody can do anything about. The two links you control completely are the light you shoot under and the white balance you set. Almost all of the correctable error is there.

The practical consequence: fixing mixed lighting and locking white balance is worth more than any amount of careful editing afterwards, because you can only correct accurately if there was one consistent light to correct for.

• Mixed light sources in the shooting space
• Auto white balance guessing differently frame to frame
• Phone processing tuned for pleasing rather than accurate colour
• Platform re-encoding and any automatic enhancement
• The buyer's uncalibrated screen, brightness and night mode

Shoot one light source, and lock the white balance

Turn things off. If you have a good window, kill the overheads and shoot by daylight. If you are shooting at night under LEDs, close the blinds so no daylight mixes in. One light colour is not a purist preference; it is what makes correction mathematically possible. With two light colours in the frame, correcting one leaves the other wrong.

Then take white balance off auto. Auto will re-evaluate for every frame, so a navy jumper filling the frame will be interpreted differently from a white shirt, and your set will drift. Set it manually and leave it for the whole session. Most phone camera apps expose this in a pro or manual mode; if yours does not, a third-party camera app will.

If you are buying one bulb for photography, buy one with a high colour rendering index and use it alone. Cheap LEDs have gaps in their spectrum, and colours that fall in those gaps — deep reds, some purples, certain teals — simply cannot be rendered accurately no matter how you set the camera.

Give yourself a truth to correct back to

Put a neutral grey or white card in one frame at the start of each setup. It does not need to be in every photograph — one reference frame per lighting setup is enough. When you edit, you set the neutral point using that card, and then apply the same correction to the rest of the set.

Without a reference you are guessing from memory, and memory is unreliable about colour specifically. Everyone who has tried to match a navy from memory has ended up somewhere between navy and slate. A card costs almost nothing and takes ten seconds.

Include the actual garment in the reference frame if you can, so you have one image where a known neutral and the fabric are lit identically. That frame becomes the ground truth for the whole listing.

Why the same navy looks different in different light

Some dyes shift dramatically depending on the spectrum of the light hitting them. This is metamerism, and it is not a camera fault — a navy that looks navy under daylight really does look black or slightly purple under certain shop LEDs, to the eye as well as the lens. Navy, charcoal, aubergine, forest green and deep teal are the usual suspects.

You cannot photograph your way out of this, because the garment genuinely looks different under different light and your buyer will unwrap it under theirs. What you can do is shoot under the light that best represents typical conditions — usually neutral daylight — and warn people in the description that the shade reads differently under warm indoor light.

Where two of your products are close in shade, photograph them together in one frame. A buyer comparing your navy against your black in the same image can see the difference far more reliably than by flipping between two separately shot listings.

Correcting a set without inventing a colour

The goal of a colour edit is to move the image towards the garment, not towards something more appealing. Set the neutral from your reference card, compare the on-screen result to the actual garment held next to the monitor, and adjust until they agree. Then stop. The temptation to nudge the saturation up because the fabric looks a bit dull on screen is how sellers end up shipping a duller garment than the photograph promised.

Consistency across a set matters as much as accuracy on any one frame. A listing page where the same black tee is warm in one image and cool in the next reads as unreliable. Foxi AI is genuinely useful here for applying the same look across a product set so the lighting and background presentation line up, with the garment's actual colour, shape and material preserved — but the correction to a true neutral still needs to be right at the source, because a consistent set that is consistently wrong is still wrong.

Keep an unedited reference file for every product. When a customer says the colour was different, you want to be able to check rather than argue.

Write what the photograph cannot carry

Describe colour in words that survive a screen. Not just green but a muted sage with a grey base. Not just red but a warm brick rather than a true pillar box. Comparisons to familiar references do more work than any adjective on its own.

State the conditions the photographs were taken in, and say plainly that screens vary. That is not a disclaimer to hide behind — you should still do the work — but it is honest, and it sets expectations before the parcel arrives rather than after.

Offer to answer colour questions and actually answer them, including with a quick extra photo under different light if someone asks. That takes a minute and it is the cheapest way to prevent a return. What nobody can tell you is by how much any of this changes your numbers; what it does is remove the errors that originate in your process rather than in the buyer's screen.

Frequently asked