Inside the lab: how we measure color before a look is released
Nothing leaves NIPIK because it looked good on one screen on one afternoon. Every camera profile is measured against reference charts, in two lights, for all 741 cameras. This is how that works, and what the numbers mean.
Why we measure at all
The human eye is a poor instrument for absolute color, and a brilliant one for comparison. Sit in a room lit by tungsten lamps and within minutes the light looks white. That is chromatic adaptation, and it is why a photographer can edit an entire reception slightly too orange and only see it the next morning.
Screens add their own error. Two monitors side by side rarely agree, and a phone agrees with neither. A color decision made by eye is made through all of that: the room, the hour, the screen, and how long you have been looking.
A measurement has none of those problems. It gives the same answer at midnight as at noon. So the foundation of every NIPIK collection, the camera profile, is built and checked by measurement, and the eye is brought in afterwards, for the part that only an eye can judge.
The test
A reference chart is a set of color samples whose exact values are known. Photograph it, and you know what every sample should look like in the finished picture. Whatever differs is the error of the camera and its profile.
- Take RAW photographs of the charts for a camera, in daylight and in tungsten light.
- Render them through the profile under test, with white balance set on the chart’s own greys.
- Measure every color sample in the rendered picture.
- Compare each one with its true value and record the difference.
- Repeat for the next camera, until all 741 are done.
The samples that matter most to us are the natural ones: skin, sky and foliage. They are the colors a viewer knows by heart, and the first ones noticed when they are wrong.
The unit: ΔE
The difference between two colors is written ΔE, said "delta E". To compute it, both colors are placed in CIELAB, a color space built around human vision, where every color has three numbers: one for lightness, one for green against red, one for blue against yellow. ΔE is the distance between the two points.
The first formula, from 1976, was a straight ruler, and an uneven one: the same number looked like a large step in some colors and a small one in others. The current formula, ΔE 2000, weighs lightness, saturation and hue separately so that one unit is close to the same visible step anywhere, for a skin tone as for a deep blue. It is the one we use.
| ΔE 2000 | What you see |
|---|---|
| Under 1 | No difference for most people, even side by side |
| 1 to 2 | A difference on close comparison |
| 2 to 5 | A difference at a glance |
| Above 5 | Two different colors |
Why one average lies
It is tempting to reduce a profile to a single score: average ΔE across the chart, lower is better. We do not, because the number hides the two things that matter.
The first is which color is wrong. An error of 3 ΔE in a saturated blue is invisible in a wedding photograph. The same 3 ΔE on skin is a face that looks ill. A profile with a fine average can be poor exactly where a photographer needs it.
The second is the direction of the error. ΔE says how far a color has moved, and nothing about where to. Skin that is 3 ΔE too green and skin that is 3 ΔE too saturated get the same score, and the first is far worse to look at. So we report results by color family, skin, greens and foliage, reds, blues and neutrals, and we look at hue and saturation separately.
Two lights, because a wedding has two
A sensor responds differently under lamps than under the sun, so a profile that is accurate in daylight can drift indoors. Every profile is therefore measured twice, in daylight and in tungsten light, and built with a calibration for each. Lightroom blends between the two according to the white balance of the photo.
Camera against camera
Accuracy is one question. Agreement is another, and for anyone who works with two bodies or a second shooter it is the one that shows in the gallery. Two profiles can each be reasonably accurate and still disagree with each other in a way the eye catches at once, because they err in different directions.
So the profiles are also measured against each other. A result counts only when it holds across cameras. A profile that wins on one body and loses on the next has solved nothing for a photographer who owns both.
What we found
- Skin lands about 31% closer to its true color than with Adobe Standard, measured across all 741 cameras.
- Foliage lands about 38% closer, and the greens stay closer under tungsten light as well.
- Between two cameras, the typical difference in daylight is under 1 ΔE, about the smallest step the eye can separate side by side.
- Greys stay neutral. A white dress stays white.
A profile is also a palette, and those numbers show where NIPIK Base spends its precision: on skin, and on the greens that surround people at almost every wedding. Skies and water are rendered calm, so they frame a couple without competing with them.
After the chart
A chart is a set of samples in even light. A wedding is thousands of colors under light that will not hold still. So the chart is where the work starts. The same profiles, and the looks built on them, are then checked on real wedding photographs, frame by frame, across whole galleries, before anything is released.
What reaches you is the result: NIPIK Base profiles for 741 camera models, with Riviera built on them. The recipe stays in the lab, the way a film maker keeps an emulsion formula. The measurements are here for anyone to read.
Questions
What is ΔE?
ΔE (delta E) is a measure of how different two colors look. Zero means identical, and about 1 is the smallest difference most people can see side by side. ΔE 2000 is the current standard formula.
What is a reference color chart?
A set of color samples with known values. Photographing it shows exactly how far a camera and its profile are from true color, sample by sample.
Do you test every camera?
Yes. Each of the 741 NIPIK Base profiles goes through the same measurement, in daylight and in tungsten light.
Is NIPIK more accurate than Adobe’s profiles?
On the colors a wedding depends on, yes: measured across all 741 cameras, skin lands about 31% and foliage about 38% closer to the true chart colors than with Adobe Standard.
Does an accurate profile replace white balance?
No. White balance and exposure are set for each photo. An accurate profile makes the color after that predictable, and the same from one camera to the next.
About NIPIK
NIPIK makes Lightroom presets for wedding and portrait photographers, together with the camera profiles they sit on.
Every collection includes NIPIK Base camera profiles for 741 camera models, calibrated to match each other. Every NIPIK Base profile is a dual-illuminant camera profile: it is calibrated for two lights, daylight and tungsten, and Lightroom blends between the two as white balance moves. Measured across all 741 cameras, skin lands about 31% and foliage about 38% closer to the true chart colors than with Adobe Standard.
The first collection is Riviera, by Danilo & Sharon: one look for the whole wedding day, in presets for daylight, flash, lamp-lit rooms and black and white, with 36 one-click add-ons, for Lightroom Classic and Camera Raw on Mac and Windows. The presets for RAW photos are ISO-adaptive: Lightroom sets their noise reduction from the ISO of each photo, leaving it at the default up to ISO 600 and raising it in steps up to ISO 25,600.
You buy a collection once. The free NIPIK Installer puts it on two of your computers and keeps it up to date. A camera that is missing is added on request. See how NIPIK compares.