Understanding Light Quality: Soft vs Hard Light, CRI, TLCI and Mixing Colour Temperatures

Last updated: October 2026

Two lights can share the same brightness, the same colour temperature and the same glowing spec sheet, and still make your face look completely different. One gives you healthy skin and soft shadows. The other gives you waxy cheeks, a green tint and shadows with edges like a knife cut. The difference is light quality, and it is three separate things sold as one: how soft the light is, how complete its spectrum is, and what temperature it runs at. This guide pulls the three apart so you can read any spec sheet, and any test clip, with clear eyes.

Light quality is three separate things

Softness is geometry. It decides whether the shadow under your chin fades gently or ends in a hard line, and whether the shine on your forehead is a glow or a hotspot. Spectrum is chemistry. Every source emits a spread of wavelengths, and gaps in that spread mean some colours in your face and clothes never get lit properly. Colour temperature is the warm to cool scale, in kelvin, that your camera tries to neutralise with white balance.

Most arguments about lighting gear go wrong because they mix the three up. A panel can throw soft light with a poor spectrum, so faces look smooth but slightly ill. A bare bulb can have a near perfect spectrum and still carve harsh shadows. Grade each part separately; the sections below take them one at a time.

Hard vs soft light: size, not price

The sun proves it. It is enormous, yet its light is brutally hard on a clear day, because from where you stand it appears as a small disc. An overcast sky becomes one giant source, and shadows nearly vanish. A 60 cm softbox one metre from your face looks huge to your skin, so shadows melt. Move it three metres back and every edge sharpens.

Diffusion on its own does not soften anything. Its job is to spread a small, fierce emitter across the full face of a larger modifier, so the modifier becomes the source. Clip it over a panel that fills only half the softbox and you still have a small source wearing a big costume. Distance matters twice: a closer soft light is softer still, and because light falls off fast up close, your background drops darker behind you, which is free separation. Placement angles and three point layouts are a craft of their own, covered in our guide to lighting your face for videos. The soft, close source is also the backbone of a flattering setup in general, as our talking head filming tips show when they build a solo shot from scratch.

Hard light is not a mistake, by the way. It is a choice. Product shots, cooking videos and anything where texture matters often look better with edge and sparkle. Faces talking to camera are where softness earns its keep, because viewers read skin before anything else in the frame.

What CRI measures, and what it leaves out

The test works like this. A lab lights standard colour samples with the lamp being measured, then with a reference source of the same temperature, and scores how far each sample drifts. The headline number, Ra, averages just the first eight samples, all muted pastels. A light can score 95 while handling saturated colours badly, because those colours were never in the average.

The famous gap is R9, the score for a saturated red sample. R9 sits outside the Ra average, and plenty of LEDs with a proud 95 CRI stamp carry an R9 under 50. Skin is full of red information: blood under the surface, lips, cheeks, the warmth that tells a viewer you are alive and well. Weak red rendering is why a high CRI light can still make people look drained. When a maker publishes the full R1 to R15 set, read R9 and R12 (saturated blue, another weak spot) before the headline.

CRI has two more blind spots. It says nothing about tint: a light can sit slightly green of neutral and still score well, because the test tracks colour samples, not the white point’s sideways drift. And it was built for eyes, which adapt constantly. Your camera records the spectrum it is given, gaps and all.

TLCI: the score built for cameras

The difference in approach matters more than the difference in numbers. TLCI runs the light’s spectrum through a model of a camera sensor and display. Because camera colour filters do not match human vision, the two scores routinely disagree. A candle is the classic example: it scores below zero on CRI, yet reaches the high 60s on TLCI, because a camera can white balance to candlelight and render a scene through it quite happily.

Broadcast reads TLCI in bands. From 85 to 100, errors are small enough that no correction is wanted. From 70 to 85, a simple pass fixes what is off. From 50 to 70, correction gets complex. Below 50, you are into rescue work that still looks wrong. For a vlogger filming skin at a desk: 90 plus is the comfort zone, 95 plus is excellent, and a maker who publishes TLCI at all usually has nothing to hide. When CRI and TLCI are both printed, weight TLCI more. Your audience watches through a sensor, not through the lab’s eyes.

TM-30 and SSI: two numbers that say more

The gamut number is the interesting half. An Rg well above 100 is pumping saturation up, which can look vivid in a demo and strange on skin. One near 100 with strong fidelity is the honest profile. The catch is availability: TM-30 appears on cinema and broadcast fixtures far more than creator panels, and SSI, developed by the Academy of Motion Picture Arts and Sciences, is rarer still. Treat both as tiebreakers: if one maker publishes TM-30 data and a spectrum graph and the other publishes adjectives, you have learned something real about both.

Colour temperature and the mixed light problem

Everyday rooms are mixed light battlegrounds. The ceiling fitting runs warm, the window pours in daylight that shifts from cool noon to golden evening, and a monitor adds its own glow. Film between a window and a warm lamp on auto white balance and you get the classic split face: one cheek tanned, the other slightly blue, the camera hunting as you move. A fixed balance stops the hunting but cannot fix both sides at once, which is why manual Kelvin control matters; our vlogging camera settings guide covers setting it deliberately.

Household bulbs make this worse because their labels lie by omission. Two bulbs both marked 2700K from different makers can differ visibly, and cheap ones often lean green as well as warm. Daylight moves too: overcast sky runs cooler than 5600K and late sun far warmer. A quoted number is a starting point, not a promise.

Mixing colour temperatures without wrecking your face

Matching is the clean solution. If your key is daylight balanced, use daylight balanced bulbs in the room’s practical lamps, or close the blinds and work fully under your own lights, or gel a warm fixture toward daylight, accepting that the gel eats some output. Switching off is the honest budget solution: a ceiling light you cannot match is one you do not use while filming. Many good looking setups are one decent key light and a room gone deliberately dark around it.

The third option is the one most polished channels quietly use: deliberate contrast. Your face sits under a neutral key. Behind you, a warm practical lamp glows amber, or cool window light rims the background. Because the two temperatures never share a surface, viewers read the difference as mood and depth, not a mistake. Mixed light only looks broken when it splits a face or a white shirt down the middle. Keep each surface loyal to one source and you can mix temperatures all day.

Why skin tones fall apart under cheap LEDs

Start with the spectrum. Most white LEDs are a blue emitter under a phosphor coating, and the spectrum commonly dips in the cyan range and runs thin in deep red. That thin red output is the R9 problem felt rather than measured: lips lose their separation from the face around them, cheeks go waxy, and skin takes a grey, tired cast. Saturation in post cannot fix it. It amplifies what was recorded, and the red information was never recorded.

Then there is tint. A light can hit its quoted temperature while leaning green or magenta off the neutral line, a drift measured as Duv that CRI does not report. Green is the common direction, and it is the fastest way to make a healthy person look seasick on camera. The third failure is instability: inexpensive lights often shift tint as they dim, and bicolour models run only half their emitters at the warm and cool extremes, so output and colour quality drop where the dial promises the widest range. None of this shows in a brightness demo. All of it shows on a face.

Reading a spec sheet like a sceptic

• Demand the pair. CRI 95 plus and TLCI 95 plus together is the working standard for face filming. One without the other is half an answer.
• Hunt for R9. A published R9 of 90 or more says the maker measured the hard part and is proud of it. Silence on R9 usually means the number is not flattering.
• Check the tolerance. Serious makers state how far a unit can drift from its quoted temperature. No tolerance stated means variation is your problem, and two matching lights may not match.
• Ask how it dims. Steady dimming that holds colour is a quality signal. A light that shifts green at low brightness betrays you in evening shoots, when vloggers use dimmers most.
• Note the light type. Daylight only models put every emitter behind one job. Bicolour models trade output at the extremes for range. Judge RGB models on their white light scores separately; colour effects say nothing about skin.

Once the numbers make sense, choosing fixtures gets easier. Our guide to the best lights for vlogging applies these scores to real panels and softboxes: this article teaches the scepticism, that one uses it.

The ten minute test footage check

The grey card clip checks tint. White balance off the card in your editor; if skin under the same light still leans green or pink, the light has a tint problem your camera cannot fix. The red object (a book cover, a piece of fruit, anything whose colour you know) checks the red rendering R9 describes: under a weak light it slides toward brown or orange. The dimmer clips check stability, because a light that drifts at half power will drift in your real evening setup too.

The side by side clip saves you later. Two lights used together must agree about what colour your face is, and units from different makers with identical quoted temperatures often do not. Film the test during the return window a seller allows. It is the only stage where your camera, your skin and your room get a vote. Specs are measured in someone else’s lab; your channel is published from yours.

Put softness first, spectrum second

None of this requires a lab coat. Read past the one big number on the box, and spend ten minutes on test clips before a light earns a place over your desk. Your viewers will never know what TLCI means. They will know your face looks right, and they will stay.

Light quality questions, answered

Keep exploring

• Best Lights for Vlogging
• How to Film Yourself: Talking Head Tips
• Vlogging Camera Settings Explained