Two lights can read the same Kelvin on a meter and render completely differently on camera. Kino Flo has spent decades on that problem, first with fluorescents and now with LEDs, and this talk is the condensed version of the colorimetry lecture he gives at universities and cinematography societies. The full video is above. The short version of the argument is below.
Why don’t two lights at the same Kelvin match?
Kelvin tells you where a light sits on the Planckian locus, the curve that white light tracks through color space. It does not tell you what the spectrum underneath looks like. Two fixtures can both read 6500K while one builds that white from a handful of narrow LED peaks and the other from a broad, continuous distribution. Deep dips in a spectrum are where color rendering falls apart, and a meter won’t show them to you. Mix several manufacturers on one set, all at the same Kelvin, and the camera sees what Kino Flo founder Frieder Hochheim calls the “dog’sbreakfast”: matching numbers, mismatched light.
There’s a second problem the spec sheets used to deny outright. LEDs shift Kelvin when you dim them. Kino Flo’s answer was an algorithm that tracks the Planckian curve through the entire dim, holding a neutral white point instead of drifting green or magenta.
Every camera is a different film stock
Film never saw light the way the eye does, which is why camera tests in prep were standard practice. Digital sensors inherited the problem. Every camera has its own spectral response, some with broader red sensitivity, some reaching into ultraviolet, some sampling in narrow bands. Working with a former senior scientist from Kodak, Kino Flo built a spectrometer that reads light through camera profiles, so the spectrum could be tailored to what sensors actually see rather than what a chart says they should.
That research also set the engineering target. Some cameras don’t track the Planckian curve. The human eye does, and cinematographers judge light with their eyes. So the eye became the goal.
How manufacturers can game a CRI number
CRI grades a light against as few as 8 color swatches. Once a manufacturer knows which swatches count, they can tune a fixture to ace the test while the spectrum still has holes a camera will find. That’s why the industry is moving to TM-30, which grades 99 swatches across the full spectral distribution. It’s a harder test to game and a far better predictor of how a light performs on set. Frieder’s position is blunt: the industry should retire CRI entirely.
What this looks like in the Celeb IKON and Diva Lux
The new Celeb IKON 6, IKON 12, and Diva Lux 4 are built on broadband phosphors tuned to camera response, with discrete RGB emitters doing one specific job: hitting the color point and holding the Planckian curve as Kelvin changes. In the latest spectra, the valley around 450 nanometers has been raised and the blue response broadened, which means more consistent color rendering not just on one camera, but across camera manufacturers on the same set.
Explore the fixtures:
- Diva Lux: https://kinoflo.com/diva/
- Celeb IKON: https://kinoflo.com/ikon/
Watch the full talk above, and subscribe to our YouTube channel for more from the Built for the Shot series: https://www.youtube.com/@KinoFloLightingSystems
