Lighting for video calls

We spend most of this site arguing that CRI is a weaker specification than the marketing around it suggests. It is an average that hides its worst case, it was last improved in 1974, and the organisation that invented it has published a statement saying it does not work properly for LEDs.

This page is the exception. On a video call, the lamp’s colour rendering stops being a sticker and starts being the image. Here is why, and what to do about it — starting with the parts that are free.

Two free fixes that beat any lamp

Before spending anything, the two changes that improve a call most cost nothing at all, because they are about where light comes from rather than how much of it there is.

Put the light in front of you, not behind you.A camera exposes for the overall frame. With a bright window behind you, the camera sees a lot of light, stops down to cope with it, and your face — lit by nothing — goes to a silhouette. Turn around so the window is in front of you and the same window becomes a large, soft, free light source landing on your face. Nothing was bought. The desk moved.

Get the source roughly at eye level or slightly above. Light from below is the classic horror-film angle for a reason; light from directly overhead drops shadows into your eye sockets. Slightly above and in front is the flattering, boring, correct answer.

There is a useful tension worth naming here. CCOHS’s guidance for screen work is to angle the monitor away from lights and windows, precisely so they do not glare off the panel. The camera wants the opposite — a window in front of you, which means behind the monitor. If you do a lot of both, you are negotiating between the two, and a window that is off to the side and slightly in front is usually the compromise that keeps everyone happy.

Why colour rendering suddenly matters

A lamp does not have to contain every wavelength to look white to you. Your visual system is remarkably good at adapting — walk from a warm room into daylight and within a minute or two, both look normal.

A camera does not do that. It records reflected light. Something can only reflect a wavelength if the lamp emitted it in the first place, so any gap in the lamp’s spectrum lands directly in the image. White balance can shift the whole picture; it cannot reconstruct a colour that was never present.

For most desk work this is irrelevant. Reading a document, you want to see black shapes on a white background; a lamp with a hole in its red spectrum still shows you the words perfectly well. That is why we treat CRI as secondary almost everywhere else on this site.

On a call, the subject is a face. Skin tone is built substantially from reds and oranges — blood under a translucent surface. A lamp that is weak in deep red gives you skin that reads grey, sallow or slightly corpse-like on camera, and no amount of fiddling in the video settings puts back what the lamp never emitted.

The number that matters is the one they hide

Here is the awkward part, and it is the same point we make on the desk lamp roundup — it just has visible consequences here.

CRI (Ra) is an average of eight low-saturation test colours, and it excludes R9. R9 is the deep-red sample. The one that governs skin tones. The one this entire page is about. It is not in the average.

So the specification that every lamp brand prints in large type is structurally silent on the only colour question that matters for a video call. A lamp can advertise Ra≥95 and still be weak exactly where your face lives, and the Ra figure is not lying — it simply was not measuring that.

This is not a fringe complaint. The CIE, the international body that definedCRI, has published a position statement saying that Ra “does not agree well with overall perceived colour rendering” for LED sources, that the issue is particularly acute for narrow-band LEDs, and that the metric needs replacing. CRI was first published in 1965 and last improved in 1974. The statement is free to read and it is in our sources.

Who actually publishes R9

Across every lamp and light bar we have specs for on this site, exactly one product publishes an R9 figure: the Dyson Solarcycle Morph, at R9 48 minimum, alongside a CRI of “90 or above.”

Notice what that combination is. Dyson publishes a lowerheadline CRI than BenQ’s Ra≥95, and is the only brand willing to show you the sample that the headline excludes. A published 90+ with a stated R9 is a more complete answer than an unpublished 95 with the difficult sample averaged out of frame. That is why it is ranked first on our lamp roundup despite the smaller sticker and the much larger price.

We should be careful about what this does and does not prove. R9 48 is not a spectacular figure — it is a floor, and specialist photo and video lighting reaches considerably higher. What we can say is narrow and specific: it is the only figure of its kind published by anyone here, and it is the only one that speaks to the question this page is asking. We have not put a camera in front of either lamp. We cannot tell you how a face looks under one, and we are not going to imply otherwise.

Worth crediting BenQ for one thing: on the ScreenBar Pro and Halo 2 it publishes Rf≥96 next to Ra≥95. Rf is the newer TM-30 fidelity index, measured across a far larger sample set than Ra’s eight. It is a different metric and must never be read as the same number — but its presence means BenQ knows the 1974 average has a problem.

Don’t mix your colour temperatures

The most common self-inflicted wound on a home call is not a lack of light. It is two kinds of light.

Your camera picks one white balance for the entire frame. If a 2700K lamp is on your left and a 6500K window is on your right, there is no setting that makes both correct. The camera commits, and the loser turns orange or blue across half your face.

The fix is matching, not maximising. If you are working next to daylight, set your lamp cool. If it is dark outside and every light in the room is warm, set your lamp warm and let the whole frame be warm together — consistent and slightly amber beats accurate on one cheek.

This is where an adjustable range earns its money, and it is a real spec difference rather than a marketing one. Dyson publishes 2700–6500K undocked, so it can meet a daylight window. The BenQ Genie tops out at 5700K, which means against a bright north-facing window it cannot get there — and BenQ’s bars, at 2700–6500K, can.

What we can’t tell you

We have not switched any of these on, we have not pointed a camera at one, and we have not compared two lamps on a face. Everything above is mechanism and published specification: how cameras record reflected light, what Ra does and does not include, and which manufacturers publish which figures.

We also are not going to tell you a lamp will make you look good. That is a claim about an outcome we have not observed, on a subject we cannot generalise about, and it is the sort of thing this site exists not to say. What we can tell you is which number governs the question, that it is deliberately absent from the metric everyone advertises, and that exactly one manufacturer here prints it anyway. Our methodology page sets out the rest of what that does and does not buy you.