Two fixtures can look identical on a showroom wall, carry the same wattage, the same beam angle and the same price, and still ruin each other the moment both are on the same stage. The reason is almost never brightness. It is that one of them renders colour differently, or that their two whites are not actually the same white.
CRI and colour temperature are the two numbers that describe this. Neither is a quality score, and neither can be read as one. This guide explains what each number actually measures, where each one misleads, and how to use both to predict what a rig will do before you hang it.
In this guide
The Colour Rendering Index is a 0–100 score that answers one narrow question: when this light falls on a set of reference colour samples, how close do those samples land to the way a reference light would have rendered them?
CRI is defined in CIE 13.3. A fixture is measured against eight standard colour samples — pastel shades of red, yellow, green, cyan, blue, violet, and two skin-adjacent tones — and each sample is scored on the colour difference between the fixture and a reference source of the same colour temperature. Those eight scores are averaged to produce Ra, the number printed on a datasheet.
The important consequence: CRI is a comparison, not a measurement of light. It tells you how well a fixture imitates a reference source of its own colour temperature. It does not tell you the light is good, bright, pleasant or appropriate. A CRI 100 fixture can be an appalling stage light.
| Sample | What it tests in practice |
|---|---|
| R1 — light greyish red | Warm skin in mid tones |
| R2 — dark greyish yellow | Golden and amber fabrics |
| R3 — strong yellow green | Foliage, and the most forgiving sample |
| R4 — moderate olive green | Warm neutral surfaces |
| R5 — light blue green | Pale cool surfaces |
| R6 — light violet | Cool skin highlights |
| R7 — light blue purple | Pale violet and lavender |
| R8 — light reddish purple | The most common failing sample on LED fixtures |
The famous gap in CRI is R9, a saturated red. It is measured as part of the extended set of 14 samples, but it is deliberately excluded from the Ra average. A fixture can score Ra 92 while R9 sits in the low teens — and saturated reds, deep crimson costumes, red gels and warm skin in shadow will then read flat and muddy. If a datasheet quotes Ra and nothing else, R9 is unknown, not good.
That is the whole of what CRI is. It is a useful filter and a poor judge. The next section covers why the number also cannot be compared across a range the way a percentage can.
CRI is often read as a percentage of quality, so a CRI 90 fixture sounds twice as good as a CRI 45 fixture and equally near-perfect as a CRI 95. Both readings are wrong.
The scale is not linear and it is not a percentage. The score is derived from colour differences, and those differences do not shrink in proportion as the score rises. The practical consequences are three.
The successor metric is TM-30, which reports two numbers instead of one: Rf (fidelity, comparable in spirit to CRI) and Rg (gamut, which describes whether colours are more or less saturated than the reference). Rg matters because a fixture can be accurate and still look dull — or slightly inaccurate and look vivid. If a datasheet offers TM-30, prefer it. Most budget fixtures do not, which is why CRI remains the practical starting point.
Colour temperature is written in kelvin, abbreviated K. The scale comes from physics: heat an ideal object and it begins to glow, first deep red, then orange, then white, then blue-white. The temperature it was heated to labels the colour it emits. A “3200 K” white is the white of an object at 3200 kelvin.
The awkward part is the vocabulary. Because the scale is inherited from heated metal, lower numbers are warmer and higher numbers are cooler. “Warm white” means around 2700–3200 K. “Cool white” means 5500 K and above. Nobody finds this intuitive on first contact, and it is the source of a great many ordering mistakes.
| Colour temperature | Reads as | Where it belongs |
|---|---|---|
| 2000–2400 K | Very warm, amber | Candlelight, period pieces, restaurant ambience |
| 2700–3000 K | Warm | Hospitality, residential, practical lamps |
| 3200 K | Tungsten anchor | Theatre, film, anything matching traditional stage lamps |
| 4000 K | Neutral | Corporate, retail, multipurpose rooms |
| 5600 K | Daylight anchor | Broadcast, film daylight balance, outdoor coverage |
| 6500 K | Cool daylight | Overcast match, industrial, high-contrast wash |
Two of those are anchors rather than choices. Almost the entire film and broadcast world is built around 3200 K (tungsten) and 5600 K (daylight). If your fixtures will ever sit alongside a camera crew’s own lights, matching one of those two is usually more valuable than picking the white you personally find prettiest.
A tunable white fixture carries both anchors. Instead of one white emitter it has a warm emitter and a cool emitter on separate console channels, which you blend to place the output anywhere between them. The practical benefit is that one fixture can be 3200 K for the play and 5600 K for the matinee recording, without gel and without a second purchase.
DT 200W COB
$169
DUALTONE 200ZM
$241
Here is where two fixtures that both say “3200 K” stop being the same fixture.
Colour temperature fixes one coordinate of a white — how warm or cool it is. It does not fix the second coordinate, which is how far the white sits above or below the line of perfect black-body colours. Two sources can both be exactly 3200 K and still be visibly different whites: one faintly green, one faintly magenta.
On its own, a faint tint is easy to miss. The problem appears when the two whites meet. Put a nominally 3200 K fixture from one batch next to a nominally 3200 K fixture from a different batch, both on the same wall, and the seam between their pools becomes obvious. On a face lit from the front by one and from the side by the other, the tint becomes a colour cast that crawls across the skin as the performer moves.
Hang the two fixtures side by side, aim both at the same matte white or light grey surface from the same distance, and set both to the same colour temperature. Photograph the wall with a phone, then look at the boundary between the two pools rather than at either pool. If you can see where one ends and the other begins, the whites differ — and no amount of extra output will settle it. Do this before you buy the second batch, not after.
The fix is inexpensive once you know which direction the difference runs. A light gel on the tinted fixture pulls it back toward the reference, and a colour-mixing fixture with an independent white channel can be walked onto the reference white from the console. What is expensive is discovering the mismatch at load-in with forty fixtures already hung.
Colour rendering matters least where the audience is a room full of eyes and most where a camera is involved, and that asymmetry surprises people.
The human visual system adapts. Give it a few minutes under a slightly green white and it will decide that white is neutral, because that is what it evolved to do. A camera sensor does not adapt; it records the ratio of light it receives and hands the file to post. Whatever tint the fixture had is now in the footage.
Three specific failures are worth knowing by name, because each one has a different fix.
If a fixture is dimming by switching its output on and off rapidly — pulse-width modulation — the camera can catch it mid-cycle. The result on screen is a band rolling through the frame, or a colour shift that appears only in slow motion. This has nothing to do with CRI, and it will not show up on a datasheet. Look for an explicit flicker-free or flicker-free dimming claim, and test it with the camera you intend to use, at the exposure you intend to use.
Most people run into colour rendering while trying to add to a rig that already exists, not while buying a whole new one. The order below is the one that avoids rework.
The venue type narrows the answer faster than anything else, because each type has a primary constraint that dominates the choice.
| Venue or use | Primary constraint | Reasonable starting point |
|---|---|---|
| Houses of worship | Fixed camera, volunteer operators, congregation readability | 3200 K or 4000 K, tunable white so the same rig serves live and stream |
| Corporate and conference | Camera, and a neutral look that suits skin and screens | 4000 K, flicker-free dimming, consistent tint across the rig |
| Theatre and dance | Matching traditional stage lamps and gel stock | 3200 K as the anchor, with cool options for atmospheric work |
| Live music and club | Saturated colour and effect, output over fidelity | Colour-mixing fixtures; treat white as an effect, not a reference |
| Weddings and events | Photography, and matching venue practicals | 2700–3200 K to agree with the room, high CRI for skin |
| Film and broadcast | The crew’s own lights and post-production expectations | 3200 K or 5600 K to match existing inventory exactly |
Read the middle column and the pattern is clear. Where a camera is involved, fidelity and flicker compact into hard requirements. Where colour is the point, fidelity matters much less than saturation and output. Buying on brightness alone ignores the constraint that will actually determine whether the rig works.
DUALTONE PAR L750
$209
LIMONA PAR 2415IP
$299
DUALTONE PAR L450
$109
They describe entirely different things. Colour temperature, in kelvin, describes how warm or cool a white looks — it is a property of the light itself. CRI describes how faithfully that light renders coloured objects by comparison with a reference source of the same colour temperature. A fixture has one colour temperature and one CRI, and neither tells you anything about the other.
Not necessarily, and the scale is not linear. Above roughly CRI 90 the improvements become small and expensive, and CRI is undefined for colour-mixing fixtures anyway — an RGBW par is not a low-CRI light, it is a light the index was not designed to describe. Use CRI to filter out fixtures that will obviously fail on camera or on skin, then judge the shortlist by eye and with a camera.
R9 is the saturated-red sample in the extended CRI test set. It is measured but deliberately excluded from the Ra average, so a fixture can score a healthy Ra while rendering reds poorly. That matters because skin, crimson costumes and red gels all live in that region. If a datasheet quotes Ra without R9, treat the red rendering as unknown.
Because colour temperature fixes only one coordinate of a white. Two sources can both be 3200 K and still differ in tint — one slightly green, one slightly magenta. The difference is invisible when they are apart and obvious when their pools meet on the same surface. Test with both fixtures on the same white wall and inspect the boundary rather than either pool.
Usually not, and the money is better spent on output and colour saturation. CRI matters where fidelity matters: skin under a camera, product photography, corporate and worship environments with a fixed camera, and theatre work that has to match traditional lamps. For a club or a band rig where saturated colour is the point, treat white as an effect and prioritise punch.
Match the camera crew rather than choosing freely. Studio and broadcast work is normally balanced at 3200 K or 5600 K, and if your fixtures land on one of those, the crew’s existing lights can join your rig without filtration. If the venue does both live and recorded work, tunable white covers both without changing the rig.
No, and that is a common and costly assumption. CRI is measured from a colour-difference test and says nothing about how the fixture dims. A high-CRI fixture can still flicker on camera if it dims by pulse-width modulation. Look for an explicit flicker-free claim and confirm it with the actual camera and exposure.
More than any other single decision after beam angle, because it determines whether the rig agrees with the room and the camera. A consistent 4000 K rig will look more professional than a mixed 3200 K and 6500 K rig of twice the output, every time — the mismatch reads as a fault, and no amount of brightness hides it.
This article describes general principles of colour rendering and colour temperature, and the published specifications of products sold on sanyilights.us at the time of writing. Colour rendering figures vary by emitter batch, mode and firmware version — always confirm the current datasheet before specifying a production. Colour temperature and CRI are engineering metrics and not guarantees of aesthetic suitability; venue requirements, camera balance and local codes take precedence. Product availability and pricing are subject to change.
