It’s easy to get into this sort of mindset because the first lights that people encounter are often small, where the latest technology works incredibly well. It’s easy to find LED lights in the 200-watt range that are more powerful than the traditional 800W redhead, beloved of TV interviewers for decades. The LEDs are small and light and can be run from batteries, at least for long enough to do an interview. What’s more, they look really bright. Fire them at the side of a building from across the street and it’s probably not achieving much by way of actual exposure, but it looks great, and that leads people to expect great things when they point a camera at it.

And, to be fair, modern cameras encourage this sort of thinking by offering us truly staggering sensitivity. There are now cameras which can literally see better than humans, and film has been left far behind. Assuming the fastest film stocks in common are or were equivalent to an ISO 500 digital camera, and assuming that most digital cameras are capable of operating at ISO 1600 with at least comparable noise, digital cameras are one and a half stops faster than film. Many digital cameras are now capable of even greater sensitivity with acceptable noise, and approaches to exposure differ, so we can reasonably credit them with a solid two stop lead.

So, lights are at least two stops better, and cameras are at least two stops better. We could even take the position that improvements in lenses have made the bottom end of the f-stop range more acceptable than the designs of decades past, with less objectionable losses in sharpness and contrast when used wide open. Even if we only accept that things are one f stop better, that still halves the amount of light we need again. Add all this up, and we’re easily looking at using one sixteenth the light we might have needed on a similar setup in the 1970s.

HMI Advantages

For some reason, though, not every 18K HMI in the world has been replaced by a 1.5K equivalent. Why? There are lots of situations in which any or all of those advantages simply might not apply.

To consider them in order, that 18K HMI we just joked about is already about as efficient as many LEDs. They’re vastly better than tungsten, but they’re not vastly better than HMI or fluorescent. Numbers indicating that LED lights are more than slightly ahead tend to refer to industrial task lighting, flashlights, work lights, or other situations where colour quality isn’t so important. Movie lights must sacrifice efficiency to achieve reasonable color rendering, and movie LEDs are often only slightly better than fluorescent or HMI.

We also assumed that digital cameras might be significantly faster than film. That’s sometimes true, although one of the world’s most popular digital cameras is often reckoned to be best used as an ISO 400 device, which makes it about as fast as Kodak’s newsgathering film that was released in the 1970s. OK, almost any modern camera has vastly more performance than that old film stock, which was designed for sensitivity and quick processing before any particularly considered imaging characteristics, but even if we’re willing to shoot it at ISO 800 it’s at best a stop, or most of a stop, faster than the sort of thing people were shooting movies on a quarter of a century ago.

4K Lens Quality

Finally, lenses. Yes, modern lenses are potentially sharper at wider apertures than older ones, but there are a number of problems with that. First, and most obviously, with modern productions often required to shoot in 4K, lenses need to be sharper, which can somewhat offset the difference. Add to that the increased demands on the focus puller on sharper formats, let alone the much more extreme demands of large-format sensors, and it’s a brave cinematographer who breaks the news to the focus puller that the whole production will be shot at f/1.3, regardless of lens performance.

Perhaps the biggest issue with glass, though, is the current fashion for lenses which are – euphemistically – characterful. The Panavision C-series 50mm anamorphic, for instance, is an adored classic. It opens up to a T2.8. Still, it was designed in a time where the bottom end of the aperture range was reserved for emergencies. Many vintage designs don’t perform particularly well wide open, making it easy to cross the line between something that’s characterful and something that’s just fuzzy. Worse, many lenses with famous idiosyncrasies only actually show those idiosyncrasies at very wide apertures where they don’t perform very well.

In short, it’s been a long time since EI 400 film was fast, HMIs were already available, f/4 was normal, but those things often haven’t changed as much as we’d like to think they have. The idea that enormously upscale shoots have been made trivial by changes in technology doesn’t quite hold water.

Grim as this all seems, there are circumstances under which new technology can make a noticeable difference. Some cameras really are capable of shooting at thousands of effective ISO without significant compromise, and some lights really do give us vastly more than any other technology ever has.

Fresnel Inefficiencies

Soft light, for instance, was traditionally created by firing a fresnel into a panel of diffusion. This is, to put it mildly, a woefully inefficient approach. Fresnel lights comprise a bulb in a tube with a lens at one end and a reflector behind the bulb. Any light that does not hit either the lens or the reflector, which is most of it, simply strikes the inside of the tube and is lost. Then, the light strikes a diffusion panel which might be made out of any number of things, most of which are functionally white objects. Some of the light makes it through and might actually get to illuminate the scene, but a good proportion of it will bounce off the way it came and be wasted.

LED can do this much better, by distributing a large number of individually tiny emitters over a large area. There are few LEDs more than a few watts per device, so all practical devices are made up of large arrays. Fitting the whole array into an area small enough to create a hard light requires complicated thermal engineering. On the other hand, a big, eight-foot-square frame designed to take diffusion can be filled with those tiny individual devices, usually at a power level much lower than watts per emitter. The result is a soft light that’s much faster to set up – there’s no light to drive it – and takes up far less space than the diffused hardlight. It’s also massively powerful, like a diffusion frame with an enormous light behind it but can often be run from a domestic wall socket.

Similarly, LED systems designed to replace things like space lights have made rigging faster and power bills lower. Replacing a 6KW tungsten-halogen space light with a 1,600W LED one saves a lot of power, which is nice. A big show, though, doesn’t use one space light. It uses dozens and replacing dozens of tungsten-halogen space lights with smaller, lighter, cheaper and more efficient LEDs saves massive amounts of money in rigging, crewing, power distribution, and extremely expensive sound stage electricity fees. That’s even nicer.

Squeezing the Middle Ground

Perhaps the takeaway from this is that those are the tools of the high end, where LED helps hugely. Those 200-watt LEDs have also made life much easier on very small shoots. The middle, as so often, is perhaps the most squeezed. In mid-2020, there are some really big LEDs that can replace HMI lights in the couple-of-kilowatt range, but they’re rare and more expensive than the similarly-efficient HMIs. It’s not easy to find really direct alternatives to things like the silver bullet Arri M18, or the 2.5K devices that can be run directly from domestic wall sockets in various parts of the world.

And none of this really addresses the elephant in the corner of the room: LED lighting is great but it is not free. A year or so ago, a very senior BSC member (who must sadly remain nameless) confided in your correspondent that the dawn of digital cinematography and high-efficiency LED lighting had not in fact saved anything on his lighting budgets. Costs had in fact gone up, not down, principally because of the swingeing cost of LEDs containing huge amounts of technology but actually emitting a pretty modest amount of light.

In short, that giant night exterior with the thousand extras is still going to require truckloads of gear and busloads of people.