Wrecks are generally one of the biggest things we try to photograph underwater. They are often enormous, and we forget that our humble little strobes are only really capable of lighting about 6ft (1.8 m) from the camera. When we're shooting wrecks we're trying to show the scale and grandeur of these bits of history, but throwing a little bit of light a long way onto these scenes has very little impact on our images, and that problem actually gets worse once we get inside a wreck and start trying to light long spaces such as corridors or rooms.

On land, nobody would light a subject the size of a bus by holding a torch above the camera and firing straight ahead. Underwater, that is exactly what most of us do when we start out, because on-camera strobes and video lights are simple, always pointed roughly where the camera is looking, and for a fish or a nudibranch a foot away, that is often good enough. Put that same light source in front of a hundred-meter wreck and the problem becomes obvious: a couple of lights glued to the camera can only light a tiny portion of what you see. It cannot show you the guns on a bow, the tanks on the deck, a motorbike deep inside a wreck. For that, the light has to leave the camera.

This is an article about doing exactly that, grounded in three wrecks that have taught me most of what I know about it, because each demands something different from the way you light it. The San Francisco Maru in Chuuk Lagoon sits deep enough that ambient light barely registers, so your lighting has to build the entire scene from nothing. The Thistlegorm in the Red Sea sits shallow enough that you are constantly balancing your lights against real daylight. And the Kensho Maru, also in Chuuk, is shallow enough to explore properly and encrusted enough that getting true colour out of its coral growth becomes the whole point. Between them, they cover most of what off-camera lighting is for.

Before any of that makes sense, it helps to be clear about the two tools in question, because they behave in genuinely different ways underwater.

A strobe fires a single, extremely short burst of light, typically around 1/500^th of a second or faster, timed to coincide with your shutter opening. Because that burst is so brief, your shutter speed has almost no bearing on how much strobe light ends up in the photo, provided you stay at or below your camera's flash sync speed, usually around 1/200 or 1/250 of a second. Whether your shutter is open for 1/60 of a second or 1/200, the strobe still only fires for its own brief duration, so the same amount of light reaches the sensor either way. What shutter speed does control, powerfully, is how much ambient light gets in around that flash: slow it down and the background blue brightens; speed it up and the background goes dark, even black, while the strobe-lit foreground barely changes. This is the trick behind the classic wreck photograph with a glowing blue background and a crisp, punchy subject in front. You are exposing for two different things with two different controls: shutter speed for the ambient background, strobe power (plus aperture and distance) for the subject.

A video light, sometimes called a continuous light, works nothing like this. It is simply on the whole time your shutter is open, exactly like the sun. That means shutter speed, aperture and ISO all affect how much of its light reaches the sensor, exactly as they affect ambient light. There is no separating them: slow your shutter down to brighten a video-lit subject and you brighten the background just as much. This is not a limitation so much as a different way of working. Video lights are controlled primarily through their own stepped brightness dial, rated in lumens, then fine-tuned through your normal exposure triangle as you would for natural light. Strobes are controlled through their own manual power dial, usually marked in stops, or through TTL metering, with shutter speed doing a completely separate job.

Get this backwards and you'll spend a long time confused about why shutter speed does nothing to your strobe-lit subject, or why your video-lit shots keep motion-blurring when you slow down to brighten them. Once it clicks, it explains almost every decision that follows on a wreck.

The San Francisco Maru sits in Chuuk Lagoon with her main deck at around 50 metres (164 ft) and parts of her holds dropping to 62 metres (203 ft), with her shallowest features cresting at around 34 metres (112 ft). At those depths, on all but the brightest days, ambient light is filtered and limited by the time it reaches the wreck; what remains is a deep blue scene with none of the intensity needed to reveal detail. She is also, notoriously, short on the soft coral and fish life that colonise Truk's shallower wrecks. However, the wreck is known as the Million Dollar Wreck: with two Type 95 light tanks still lashed to her forward deck, trucks, sea mines, torpedoes and stacked ammunition crates sitting largely undisturbed, nothing here softens the hard edges of the cargo, and nothing distracts from getting the lighting right.

This is technical diving territory, well beyond recreational limits, and that changes how you think about lighting before you even leave the surface. Bottom time is short, task-loading is high, and you won't get many chances to swim around a subject repositioning lights on the fly the way you might on a shallow reef. The lighting plan for a shot of one of those tanks needs to exist in your head, ideally sketched out during the dive briefing, before you're anywhere near 50 metres (164 ft).

What tends to work is treating the scene the way a studio photographer treats a still life: multiple lights doing different things. I like to use a couple of static video lights to light a key feature of the wreck — this can be the gun on the bow, or the tanks on the deck — then use a model holding another light source that can be aimed at the subject to highlight it even more. Video lights are especially popular for this kind of diving, since a continuous beam lets you and a buddy see where the light is falling before committing to a shot. This is especially important when time is of the essence: it takes much longer to set up strobe lights, since you have to fire repeated test shots to see exactly where the light falls and then adjust accordingly.

The Thistlegorm could not be a more different lighting problem, which is precisely why it's worth discussing alongside the San Francisco Maru. She sits with her deck around 17 metres (56 ft) and her keel closer to 30 metres (98 ft), shallow enough that on a clear Red Sea day there is real, usable ambient light throughout the dive, sunlight raking down and picking out the wreck's silhouette against a genuinely blue background.

For me though, the Thistlegorm lighting challenge comes to life once you head inside the wreck. The cargo hold sits around 22 metres (72 ft) on the upper deck and 26 metres (85 ft) on the lower deck, and inside it's filled with a whole host of goodies — trucks, motorbikes, munitions and more. The cargo hold is exposed to the ambient light around the edge, so if you position yourself pointing towards this opening, you're shooting from a dark space into ambient light. This is great, as it allows you to use slow shutter speeds to burn in the blue for your backgrounds, which adds much more depth to the scenes you're shooting.

Once in hold two, I like to use on-camera strobes, off-camera strobes and video lights together — the more the better. On-camera strobes are perfect for the foreground, easy to adjust and make the scene nice and sharp.

You can then slow down the shutter speed to add the blue into the background. Once your exposure is set, it's time to start filling the gaps with your other light sources. Off-camera strobes work great inside the cabins of the trucks, and you can just use their power settings to control the brightness, usually leaving higher power on those that are further away from the camera.

On my recent trip last month, I used children's swimming floaties to float video lights into the rafters of the wreck. This allowed me to light the scene from above rather than across it, like usual. I was very pleased with this effect and will continue to play with it as I dive more and more wrecks.

The Thistlegorm's popularity brings another problem: heavy traffic. The wreck is popular, and you are often waiting for divers to pass through the scene before you can continue with your shot. Usually I just hide myself in the dark corners of the wreck out of the way while I allow other divers to pass through the scene. This works well, especially if you're only using remote strobes, as they're not emitting any light — but once you use video lights that are constantly on, you can become quite frustrating for other divers, so be careful of what you're doing, and remember other people have just as much right to be there and enjoy the wreck as you do. Traffic in the wreck is generally one-directional, managed by local guides, so make sure any video lights you use shine the opposite way to the traffic flow, so you're not shining high-powered lights into other divers' eyes.

The Kensho Maru sits at a friendlier depth than her Chuuk Lagoon neighbour: superstructure around 18 metres (59 ft), deck near 24 metres (79 ft), hull settling on the bottom between roughly 32 and 40 metres (105 and 131 ft) — shallow enough to explore properly on a single recreational dive and linger in her genuinely excellent engine room and pilothouse. Decades in warm, sunlit water have done what they always do here: soft coral and sponge growth now cover much of her superstructure.

In the engine room, there is the most incredible skylight that you can use for the top of your frame, allowing ambient blue light to light the top, then you use your off-camera lighting and a model to light the fine details in the shadows of the engine room — the panels, the engine itself and all the fine historical artefacts that are there. You can get as creative as you like here, but I find video lights work great in these tight, dark spaces, as they immediately allow you to see what you're doing, what cool things there are to light, and let you manipulate the scene much more quickly.

Whatever the wreck, the mechanics of taking lights off the camera are the same, and it comes down to two questions: how do you trigger them, and where do you put them.

Strobes fire either through an electrical sync cord, reliable but limiting how far the strobe can travel from the camera, or, more commonly now, through fibre-optic cables, which use a sensor that picks up when your on-camera flash fires, then sends that signal down to the off-camera strobe. Basically, you can hide the strobe anywhere you want, then run the cable out, and as long as the sensor is in the eyeline of your on-camera flash when it fires, it will trigger the remote strobe.

Fibre-optic triggering is what makes handing a strobe to a buddy workable, since there is no cable tethering it back, only a slim optical line leading to the sensor. Once triggered, a strobe's brightness is set on the unit itself, using the manual power dial. I often just leave this on full power, especially if it's far from the camera, then use camera settings to adjust the fine details.

Distance matters more than most expect: intensity falls off with the square of distance, so a strobe brought twice as close delivers four times the light, which is why a modest repositioning changes exposure more than nudging the power dial by a stop.

Video lights need no triggering at all, since they are simply switched on before you descend, but they demand more discipline in exposure because, as covered above, their contribution is inseparable from shutter speed, aperture and ISO. The practical routine is to pick your ambient exposure first — shutter speed for the background, aperture for depth of field — then bring the video light's brightness dial up or down until the subject sits correctly within that exposure, checking the result on-screen rather than assuming. Because the beam is continuous, a slow shutter speed that would freeze a strobe-lit subject perfectly can motion-blur a video-lit one if your buddy's hand, or the subject, moves during that longer exposure.

What I'm planning to shoot determines how I rig my lights. Often they're on GorillaPods with no flotation, which means they hold pretty steady in the water and I can change their angle very easily.

As discussed earlier, I'll also bring floats so I can float lights up into ceilings; the head of the light, which is the heaviest part, then points directly down, lighting my scene.

When positioning your lights, never point them at the camera — they'll just blow out the whole shot and cause lens flare. Instead, use them to light key features and backgrounds to create depth, always pointing them away from the camera. There is one exception to this: backlighting a diver or a feature. In that case, put the light directly behind the model or subject, pointing at the camera. This will create a sunburst effect of light around the subject, a technique commonly used in cave photography these days.

None of this is complicated in principle, only in execution, because it asks you to do something that feels unnatural at first: stop treating the light as an extension of the camera and start treating it as an independent tool placed in the scene, the way you'd position a lamp in a room rather than strap a torch to your forehead. The San Francisco Maru forces you to build an entire scene out of nothing but the lights you brought down. The Thistlegorm hands you real daylight and asks you to control two exposures at once: blue water through your shutter, wreck through your strobe.

Every wreck comes with new lighting problems, but in every case the answer is the same: get the light off the camera, and let it light the parts of the wreck you can't, highlighting the features and points of interest. It's a small, deliberate step to take before a dive, and it's the difference between a photograph that documents you were there, and one that shows what you actually saw.