Production, Video

Video light challenges: CRI vs. TLCI-2012 and three LED lights


For video lighting, we are always looking for the most accurate colour rendering. In fact, what we are looking for in a video light is an equivalent of sunlight or incandescent light at an output that is high enough to overpower the noise levels of the camera sensor. When plotted on a frequency chart, the light emitted by the sun or an incandescent light bulb shows as a smooth graph representing all the colours we can see. An increasing number of video lights are based on LED technology. The frequency chart of some LEDs — especially cheap ones as well as the so-called ‘white LEDs’ — shows they can’t render deep reds and blues well. This obviously reduces their usage for video production. The CRI index doesn’t help much as it’s most commonly used to help the marketing department. So, what should you be using instead? And from the three LED lights I reviewed — Akurat Lighting LL2120hc3, Akurat Lighting A1 and BALLED Pro — which are the best for video and/or photography?

Because sunlight and incandescent light show a nicely balanced output in all frequencies of the visible spectrum, people with full vision capacities can differentiate all the known colours. Equally so, this balanced output means a digital camera sensor or film emulsion will register colours bouncing off objects faithfully.

Even if the light you’re using isn’t perfect, your eyes have the ability to adapt to its characteristics and with the help of ‘memory colours’ you will still be able to see objects and subjects with a relatively faithful representation of the colours the materials reflect when the sun shines on them.

However, this does not apply to cameras or film emulsions. These need to be profiled or calibrated to the colour temperature of the light you’re dealing with before colour rendition is acceptable.

White light or rather the impression of white light can be obtained by mixing two or more lights with complementary colours. The spectrum of such light is the sum of the spectral components creating the impression of white light. Most often this is not a continuous spectrum and examples of lights that behave as such are white fluorescent lamps and… white LEDs.

To use white LEDs as video lights they have to be adjusted to the same colour temperature as the natural sources. However, the registered colours may differ from the natural ones due to the incomplete or distorted spectrum they output. White LEDs may generate distorted hues, represent a known colour different than the natural one, or completely lack the registration of some colours. Especially when a video light lacks colours, adjusting the camera or using correction filters will not help — what isn’t there can’t be made to appear from scratch.

The Colour Rendering Index (CRI) and why it serves marketing so well

In order to help us evaluate the difference between the spectrum of a specific light source and the spectrum of natural light, the CIE created the Colour Rendering Index (CRI).

The CIE based the CRI parameter upon its tried concept of the “standard observer” and how the human eye perceives colours. The index requires creating a special table of 14 sample colours (Test Color Samples or TCS) that are called TCS1 to TCS14. Simply put, the CRI depends on a “perfect” light source, which is a black body radiator for sources with correlated colour temperatures under 5000 K, and a phase of daylight otherwise (e.g. D65).

The way the evaluation process works is by illuminating TCS1 to TCS14 colour patches with a test light and then finding out how similar the colour rendition of each patch is to the reference colour. This process results in a series of colour rendering indexes (R1 to R14).

The R1 to R14 indexes for the reference light are equal to 100. The larger the difference between the reflections of the tested light and the reflections of the reference light for the particular TCS colour, the lower the corresponding index will be.

Light manufacturers will usually publish an “Ra” index, which is the arithmetic mean of the first eight indexes — R1 through R8. As such Ra does not reflect the entire index range. Not surprisingly, two of the indexes left out — R9 and R12 — represent the outermost regions of the spectrum. They correspond to the deep reds and blues most white LEDs can’t reproduce.

Depending on Ra to describe the spectral quality of LEDs is therefore a bad idea. Worse yet for video and filmmaking use, Ra describes neither colour temperature nor green–purple balance, and those are fundamental to capturing colour images with video cameras.

TLCI-2012 to the rescue for LED video light

In 2012 the European Broadcasting Union (EBU) approved a new standard of quality evaluation for video lights. TLCI-2012, (Television Lighting Consistency Index 2012) first defined “standard camera” as a parameter that is different from the CIE’s “standard observer”.

The new standard is based on the spectral characteristics of a light source measured using a spectrophotometer. A TLCI index of 100 means there is no need for colour adjustments in terms of uniformity. Indexes lower than 100 imply some adjustments will need to be made. The lower the TLCI, the more the light deviates from the reference values, both in colour balance and spectral curve.

The TLCI-2012 standard is the preferred index to evaluate video lights without having to measure them yourself, but it is criticised by some. It’s also not meant to be a standard ‘carved in stone’. The EBU website itself states it’s a “Recommendation designed to give technical aid to broadcasters who intend to assess new lighting equipment or to re-assess the colorimetric quality of lighting in their television production environment.”

Most criticism that has been formulated against TLCI-2012 is related to the “standard camera”. Some have claimed the standard to be biased. A Lead Research Engineer and expert on colorimetry, displays and future TV systems at the BBC has argued this claim to be false, the standard not to be biased at all and explained why the TLCI-2012 standard is the best way to compare video and other production lights in terms of output accuracy.

How do Akurat Lighting, BALLED Pro and Relio video lights stack up?

Of the three LED lights I reviewed, Akurat Lighting’s A1 on-camera LED light has a CRI as well as a TLCI-2012 of 98. Its older LL2120hc3 comes with a CRI of 98 and a TLCI-2012 of 99. BALLED Pro only mentions the LED’s colour temperature. You can find an extensive review of the Akurat Lighting V-WHITE LL2120hc3 and A1 on-camera video lights respectively on this site, but no tests with regards to colour accuracy were illustrated. The BALLED Pros have been reviewed too.

One range of lights that I haven’t yet had a chance to review are Relio’s lights, a range of Italian-made, beautifully designed small LED lights. They do announce their respective models’ CRI indexes, which fall short of the Akurat Lighting products at an average of 92 to 96 (depending on the colour temperature of the three lights they make as you can read from their technical specs — yes, they publish those!). It’s important to realise Relios aren’t marketed as video lights, but as photographic lights, just like the BALLED Pros.

I did some testing of my own. I’m not going to claim my tests are scientifically correct, but I think they do give you a good impression what each light is capable of (correct me if I’m wrong and tell me why). For measuring I used DSC Labs’ ChromaMatch Pro, a colour reference chart, each of which is individually ran through quality control (spectrophotometer) multiple times before it leaves the factory.

These charts are used in the film industry and are the most accurate available. I reviewed the ChromaMatch Pro here. After white balancing the camera, I took a shot of the ChromaMatch, cropped the frame and loaded it in Final Cut Pro X because its vectorscope is one of the easiest to read in my opinion.


I wonder which CRI daylight would have…

I first shot the ChromaMatch in daylight. This served as a reference. You’ll notice that it was a bit overcast as none of the colours reach the target boxes of the vectorscope. I had a bit of reflection that I could not get rid of as well. That shows in the scope as the “shoots” in the middle of the scope. Note that the green is a bit off axis.

BALLED Pro does not publish its CRI.

BALLED Pro does not publish its CRI.

You’ll note that BALLED Pros spike to the target boxes at each of the six vector colours (red, green, blue, magenta, cyan and yellow), with cyan overshooting its target and green missing its target in much the same way as the daylight shot.

Akurat lighting LED video light

Akurat Lighting’s LL2120hc3 has a CRI of 98 and a TLCI-2012 of 99.

The Akurat Lighting lights show the same, but in a less spikey arrangement.

video light A1

Akurat Lighting’s A1 has a CRI of 98 and a TLCI-2012 of 98.

What these vectorscopes show me is that all three lights are capable of rendering blue well, while the Akurat lights are more balanced in their output than the BALLED Pros, but all three are fairly close to a vectorscope image of an overcast day.