If we didn’t care about efficiency, tungsten lighting is the best technology ever developed. It has near perfect colour rendering, it’s lightweight, inexpensive and dimmable. It’s great. Sadly, we do have to care about efficiency, although instead of efficiency we should really talk about luminous efficacy, since the term takes account of the way the human eye works. After all, if we could see long-wave infra-red radiation, the heat of tungsten lights would suddenly seem a lot more useful. Even once we’ve got a number that represents the amount of light coming out of something, though, it’s not always straightforward to make comparisons.
The issue is that lots of the more efficient lighting technologies, particularly LEDs, have a very different physical layout to things like basic tungsten lightbulbs, and that can alter how effective they are, for good or bad. This wasn’t so true with HMI, which was developed in the 1960s. It’s more or less a one-to-one replacement for tugsten. It’s a glass envelope from which light emerges in more or less all directions. Replace a tungsten bulb with an HMI bulb of the same wattage, and you have roughly quadrupled the efficiency of your light, and nothing else changes. The light operates the same, and light comes out of it in the same way; it just consumes less power for the same exposure. That’s not only useful, it’s also really easy to understand, and really easy to promote.
Many other light-generating technologies aren’t glass bulbs with a small, intense point of light inside. We can measure the number of lumens per watt that come out of a fluorescent tube (seventy or so, for movie types) and compare that to what tungsten halogen can do (twenty, maybe) and note that the fluorescent tube is vastly more efficient. The problem is, we can’t make a fresnel light that’s driven by a fluorescent tube because it isn’t a sufficiently concentrated source. So, depending on what sort of light you actually need, the comparison isn’t very vrelevant.
To fully understand what’s going on with any particular light, we need to think about the whole system, and that’s something that isn’t often done. Consider, for instance, a big soft light. Traditionally, that might be a large diffusion frame with a big Fresnel behind it. A Fresnel is a design that trades off luminous efficacy for convenience. In a traditional design has a reflector behind the bulb, a condenser lens in front of it, and the Fresnel lens itself at the front. The bulb emits in all directions. Any light that doesn’t hit either the condenser or the reflector is blasted into the housing and lost. Light that makes it through the lens hits the diffusion, which is substantially a white surface, so a lot of that bounces back and is also lost. Anything that makes it through the diffusion may, give or take flags and cutters, go on to illuminate the scene.
Compare that to a textile with LEDs woven into it. The LEDs are vastly more efficient than tungsten-halogen, although the truth is they’re probably around the same efficiency as an HMI. The real difference is that there’s no reflector, no condensing lens, no fresnel, no diffusion. It’s massively more efficient simply because there’s so much less between the light source and the thing it’s supposed to be illuminating. That probably makes more difference than the improved efficacy of the light source.
That’s a pretty extreme example, but it should be clear that it’s very difficult to compare technologies based on numbers as fundamental as luminous efficacy. A more realistic instance might be common small LEDs such as the Aputure COB series, NanLite Forza, or Godox SL. They’re modelled broadly after the flash heads used by stills photographers, but the layout of the emitter and reflector is probably closest to an open-faced light, something like a redhead in continuous-lighting terms. In that situation, the improved performance of the LED over the tungsten-halogen lightbulb is clearer, though for a 100-watt Godox to equal an 800W tungsten redhead, the efficacy improvement would have to be 8:1, whereas in reality it’s probably nearer 4:1. Things improve when manufacturers begin to pay attention to detail; Aputure released a highly-polished, multifaceted reflector for their COB series which improves output by at least a stop.
Wait, what – we can have an extra stop just for a better reflector? Yes, we can. The redesign turned the COB-300D Mk. II into something more like a PAR, with the attendant decreases in shadow quality and beam purity – there are reasons stills people call more diffused reflectors a “beauty dish” – but it pushes out a whole lot more photons if that’s what you want.
You can also turn most of those little COB LEDs into a fresnel with a Bowens-mount adaptor, although, again, details matter. A Zylight F8 has, as the name suggests, an eight-inch lens. Many of the Bowens-mount Fresnel adaptors are five-inch, and only the light that happens to come out of the LED and hit the lens makes it out of the front. The bigger the lens, the brighter the beam. Realising this, Aputure went on to design an improved Fresnel adaptor with a larger main lens and an integral condenser, like a traditional Fresnel, to direct more of the LED’s output in a useful direction. The result, again, was a whole extra stop.
Optics, it seems, matter. This is an argument that’s been going on since the dawn of cinema. Silent movies were very aware of power consumption, and experimented with Cooper-Hewitt mercury lighting which operated very broadly like a fluorescent tube. Photos of sets from 1915 seem to feature time-travellers using Kino-Flo Vista Beams, which explains where at least some early commercial filmmaking got its big soft lights. As such, any time someone tells you an LED light is x times brighter than some other technology, be aware that ever since 1915, the answer has been that it depends.