Product Photo Color Accuracy: Where Color Actually Breaks

Product photo color accuracy breaks in five specific places between your studio and your buyer's screen, and only four of them are yours to fix. Suppliers usually attack the wrong one — they buy a better camera when the damage was done by an export setting, or they re-shoot the whole range when the real problem was a fluorescent tube above the shooting table.

This matters more for exporters than for anyone else. A domestic buyer who thinks the walnut is too red drives to the showroom. An overseas buyer opens a dispute, and "colour not as shown" is one of the hardest claims to defend, because your photograph is the evidence and it disagrees with you.

Here is the chain, in order, with the failure mode at each link and what it costs to fix.

Handoff 1: The light above the product

Every colour you will ever photograph is a conversation between a surface and a light source. Change the light, change the colour — and not proportionally, which is the part that catches people out.

Two samples that match perfectly under your studio light can visibly differ under a shop's LED panel. That is metamerism: a colour match that holds under one illuminant and fails under another. It is not a defect in the dye or the coating, and no amount of photography fixes it. It is why the paint industry runs visual comparisons inside a controlled booth under agreed illumination — the method is standardised in ISO 3668 — and why textile colour change is graded against the ISO 105-A02 grey scale rather than by eye in whatever light is available.

What to actually do:

Shoot under a single light type. Mixed daylight-plus-tungsten is the most common cause of a colour you cannot correct later, because the two halves of the product need different corrections.
Use lamps with a high colour rendering index. Cheap lighting with a low CRI has gaps in its spectrum, and a colour that lives in one of those gaps will photograph wrong no matter what you do downstream.
Kill the coloured bounce. A red brick wall two metres away, or a green cutting mat under the product, tints everything. Neutral grey or white surroundings only.

Handoff 2: The camera's guess

Auto white balance is a guess made by an algorithm that has never seen your product. On a shot dominated by one colour — a full frame of teak, a whole pallet of blue drums — that guess is reliably wrong, because the camera assumes the scene should average out to neutral and pulls the dominant colour toward grey.

The fix costs about the price of lunch: put a neutral grey reference card in the first frame of every setup, then remove it and shoot the set. In editing, click the grey card to set white balance, and apply that same correction to every frame from that setup. That single habit removes most colour drift between SKUs shot on different days, which is the drift buyers notice when they see two of your products side by side in a catalogue.

Shoot raw if your camera offers it. White balance in a raw file is a setting you can change afterwards with no penalty; in a JPEG it is baked into the pixels.

Handoff 3: The screen you edit on

If your monitor is uncalibrated, you are correcting colour toward whatever that panel happens to be wrong about. Every correction you make is inverted on a correct screen.

This is where standards earn their keep. ISO 3664 specifies the viewing conditions used for critical appraisal of images — the illumination, the white point, the surround — precisely because "looks right to me, here, now" is not a transferable statement. You do not need a full standards-compliant booth to sell furniture, but you do need two things: a calibrated display, and a habit of judging colour in a room whose light doesn't change between morning and afternoon.

The measure of "how far off" is Delta E — a single number for the perceived distance between two colours, most reliably calculated with the CIEDE2000 formula. Roughly speaking, a Delta E near 1 is about the smallest difference a trained observer reliably picks out when two samples sit side by side under controlled light; differences well above that are the ones a buyer notices on a screen without any reference at all. Treat it as the unit the argument is conducted in, not as a pass mark you can print.

Handoff 4: The export — where most damage actually happens

This is the link that silently ruins more listings than the other four combined, and it is pure settings.

The web assumes sRGB, the default RGB colour space defined in IEC 61966-2-1. Browsers and most marketplace pipelines treat an untagged image as sRGB. So if you edit in Adobe RGB — a wider gamut, genuinely useful for print — and then export without embedding the profile, every saturated colour gets reinterpreted into a narrower space and lands flatter and duller than what you approved. Reds go brick. Teals go grey.

Export mistake	What the buyer sees	Fix
Adobe RGB exported untagged	Dull, desaturated, especially reds and cyans	Convert to sRGB on export
Profile stripped by a compression tool	Same desaturation, appears after the image "worked"	Re-check colour after every compression step
CMYK file from the print catalogue reused online	Muddy, sometimes colour-shifted entirely	Keep a separate sRGB web master
Screenshot of a PDF used as a listing image	Double conversion plus compression artefacts	Export from the original, never from a document

A useful discipline: after every automated step in your pipeline — resizing, compression, marketplace upload — open the result and compare it against the master. Most teams check the first export and never check what the pipeline did afterwards.

Handoff 5: The buyer's screen, which you do not control

You cannot calibrate a purchasing manager's laptop in Rotterdam. You cannot control whether they are viewing it under a window at noon or a desk lamp at midnight. This link is permanently, structurally out of reach.

Which is exactly why the colour has to be stated as a code, not only shown as a picture. A photograph is a persuasion tool; a colour code is a contract term. RAL for powder coatings and paints, Pantone or a lab-measured reference for textiles and plastics, a named wood finish with a physical sample number for furniture — these are the things that survive a screen you have never seen.

The rule that ends most colour disputes before they start: show the colour in the photograph, name the colour in the spec, and ship a physical sample for anything above a threshold order value. Every one of those three is doing a different job, and none of them substitutes for the others. When a claim does arrive, the difference between a defensible position and an expensive one is usually whether a code was ever stated in writing — the same evidential logic that decides goods not as described claims.

Product photo color accuracy: the five-link summary

Link	Failure mode	Cost to fix	Who owns it
Studio light	Metamerism, mixed sources, colour bounce	Low — lamps and a tidy set	You
Camera	Auto white balance guessing wrong	Very low — a grey card	You
Editing screen	Correcting toward a wrong panel	Medium — calibration device	You
Export	sRGB conversion and stripped profiles	Zero — settings only	You
Buyer's screen	Uncontrollable variation	Not fixable — anchor with a code	Nobody

Four of five are cheap. The expensive-looking one, the monitor, is the only real spend on the list, and it is still less than one returned container of the wrong-coloured chairs.

Next steps

If you sell finishes, colourways, or material options, work down this in order — it is sequenced by cost per unit of improvement, not by how impressive it sounds:

Fix the export first. Confirm every listing image is sRGB. This is free and takes an afternoon, and for many suppliers it is the entire problem.
Add a grey card to every setup. Cheap, immediate, and it makes your whole catalogue internally consistent.
Clean up the lighting. One light type, high CRI, neutral surroundings.
Calibrate the editing display. Now your corrections mean something.
Put the colour code on the image itself. This is the step most suppliers skip, and the one that removes the most email.

That last step deserves its own sentence, because it is where photography stops and specification starts. A finish swatch photographed beautifully still leaves the buyer guessing which of your four walnut tones they are looking at. The same image with the finish name, the RAL or Pantone reference, and the measured dimensions of the visible panel called out directly on it answers the question in any language, on any uncalibrated screen. The important constraint is that those labels must come from real measured and recorded values pinned to the right region of the image — annotation software that snaps to the actual edges of the product and exports at each marketplace's required size does this reliably, whereas an AI image generator will cheerfully restyle your walnut into something warmer and attach a colour name it invented. A restyled photo with a plausible label is precisely the artefact that creates the dispute you were trying to prevent. The practical layout patterns are in show finish and material options, and the annotation approaches themselves are compared across tools to add dimensions to product photos.

FAQ

Why do product colors look different online than in real life?

Because at least five independent things changed between the product and the screen: the light it was photographed under, the camera's white balance decision, the monitor it was corrected on, the colour space it was exported in, and the buyer's own display and room light. Product photo color accuracy is a chain, not a setting — a perfect camera cannot rescue an image that was exported out of Adobe RGB without a profile, and a perfect export cannot rescue a shot taken under mixed lighting.

What Delta E is acceptable for product photography?

There is no universal pass mark, and any supplier quoting one as an industry rule is inventing it. Delta E measures the perceived distance between two colours, best calculated with the CIEDE2000 formula, and a value around 1 approximates the smallest difference a trained observer reliably detects side by side under controlled viewing. Buyers viewing a screen with no reference sample are far less sensitive than that — which is why the practical tolerance for a listing image comes from your category and your return history, not from a number copied off a monitor review.

Should I use sRGB or Adobe RGB for product photos?

sRGB for anything going online, without exception. sRGB is the default colour space browsers and marketplaces assume, defined in IEC 61966-2-1, so an Adobe RGB file exported without an embedded profile gets reinterpreted and comes out visibly duller. Adobe RGB has a genuine role in print catalogues where the press can reach those colours — keep it as a separate master and convert on export for web.

How do I keep colors consistent across products shot on different days?

Photograph a neutral grey card in the first frame of every setup and use it to set white balance for that whole batch, keep the same lamps and the same room, and correct on a calibrated display. Consistency between your own SKUs matters more commercially than absolute accuracy, because a buyer comparing two of your products in one catalogue spread will notice a mismatch between them long before they notice that both are slightly off the physical sample.

Does stating a RAL or Pantone code actually reduce disputes?

Yes, because it converts an opinion into a checkable fact. A photograph is evidence of appearance under unknown conditions; a colour code is a term the buyer accepted at order time and either matches or does not. For coatings, RAL; for textiles and plastics, Pantone or a measured lab reference plus a physical sample. Stating the code on the spec sheet and repeating it on the product image itself is what keeps a colour conversation from becoming a colour claim.

Product Photo Color Accuracy: 5 Places the Color Breaks