Some problems and solutions for digital underwater photography
Guido Zsilavecz
Cape Town, October 2003


Introduction

Digital cameras were quickly adapted for underwater usage by their and third-party manufacturers; much quicker, in fact, than film cameras of equivalent size and functionality were, even though they have been around for much longer. The popularity of digital cameras has increased dramatically over the last few years. While still not cheap photographers, confident in the knowledge that there would not be any further expenses on development, flocked to digital underwater photography. Partly this also had to do with the perceived complexity of Nikon's Nikonos V, or even Sea&Sea's Motomarine MXII, while ostensibly offering better quality results than, say, Sea&Sea's MX10. In many ways the perceptions are true, but the reality is that only lately have digital cameras started offering photographers cameras of equal potential as similarly priced single-lens reflex (SLR) cameras - until then, and still in many cases, they are nothing but digital happy-snappy's - the PhD (push here dummy) cameras for holiday memories, easy to use, acceptable quality, but nothing for National Geographic. For the serious underwater photographer the cameras simply did not offer anything near the abilities of the Nikonos V or a housed Canon EOS50e or Nikon F90, or only at astronomical prices!

Without really understanding the problems many divers have bought digital cameras and their associated housings, only to be disappointed by the results. This is partially simply due to the photographers lack of knowledge of photography, but also because of the camera's limited abilities. In this article I will attempt to explain some of the problems and show some solutions.

It is all about light

On land flash photography is a nice-to-have when the sun sets and the party starts. Under water a flash is used for I would guess 99% of all images. The absorption of the different colours with depth (and distance too!), and the lack of light make the use of a flash mandatory. The main problem with underwater digital photography (and technically, this applies equally with any happy-snappy style camera underwater), is the ability to both provide and control the required light.

To explain this topic of providing and controlling light let me first discuss both existing (film) underwater and land cameras, their flashes, and the way they do things, before showing how this relates to digital cameras.

Nikonos, Sea&Sea and other underwater cameras

The standard bench-mark for underwater photography is Nikon's Nikonos system. The Nikonos V, now no longer produced, is a solid, robust, simple but very efficient camera, which together with its flashes is still the tool of choice for many underwater photographers. The camera evolved from the fully mechanical early models to the semi-automatic IVa (which was not a good camera at all, do not buy one!), and the much improved V. During the V's life-time it was coupled with several flashes: the large and very powerful SB102, the smaller SB103, and its improved successor, the SB105. What made the camera and flashes so formidable was the TTL (through the lens) control of the flash. The method is simple: trigger the flash, wait until enough light has reached the film plane to expose correctly, then switch it off. To assist the electronics a small table showing which aperture to use for which distance, given the film speed, was provided - it allowed the photographer to preset more or less, while the TTL did the fine-tuning. Yes, it was possible to over-expose, especially if the subject was small and surrounded by water (think silvery fish), or if the subject was on white sand. Similarly, it was possible to underexpose, with the flash having fired fully. Thankfully the flash tells you so. To compensate the photographer could change aperture, or in some circumstances "fool" the camera by changing the program through setting a different film speed - a lower film speed to underexpose, a higher one to over-expose. Simple, crude but remarkably effective.

This simple model was carried over by Sea&Sea and their cameras, offering a similar table, slightly less control (film-speed could not be set as fine as on the Nikonos, for example), but all in all a very decent package. Housed cameras either used Nikonos flashes and a similar method, or housed flashes and whatever camera's fancy flash control system, with much less user input required.

A slightly different model was used by the Nikonos SB101 and IVa flash and camera. The IVa did not have TTL, but the SB101 compensated by having an external TTL, as it were: a small light meter which would shut off the flash as required. Sea&Sea's upgraded MX10 provided pretty much the same idea: set the camera for a specific distance to a set aperture, and the flash would take care of the rest. The results were more error-prone than proper TTL, but all in all the system gave decent enough results most of the time.

Land cameras and flashes

It has always surprised me that for land cameras the exposure table the Nikonos flash came with was not available - I have yet to find one for my Canon EX380 flash. There is certainly nothing available for built-in flashes found on the many mid-range SLR cameras. This is a great pity, because for all their ability to meter in multiple segments, centre-weighted and more, the cameras behave in a rather dumb way when it is dark. I must here admit that I know Canon's products best, but have seen similar behaviour in other manufacturers products, and that this is not an attack on Canon - I actually like their products otherwise.

What the exposure table essentially does is allow you to pre-set the exposure, given, as mentioned, the distance to subject and film speed. What this actually means is that the closer you are, the more you can close the aperture, and doing that increases the depth of field, which most of the time you really want. Unfortunately all my Canon land cameras do is look at the ambient light, and open the aperture. This is technically sound: you need to open the aperture to let more light in, except for two rather important factors:

  • The flash is on, and the camera knows it
  • The camera knows the distance to subject - the auto focus tells it
So why would it open the aperture? If I am sitting a meter from a subject, even if it is pitch black, surely f2.8 is not required, and we can probably close down to f16? No, the camera does not seem to believe so, and as there is no table, any setting I want to use to override would be guess work (or rather, by now, experience.) Thankfully the flash tells me whether I have exposed correctly or not, and can thus adjust accordingly. Of course, I can just let the camera's TTL take over and fire away, except that I simply do not want every photo to be on f2.8 without depth of field!

Why is the above a problem? Because digital cameras follow exactly the same model.

The case of white pictures

A few years back when digital was still relatively new and underwater usage was limited, I received an e-mail from a photographer complaining about white pictures. Turned out they were simply over-exposed, but the photographer could not understand why, given that camera, housing and external flash were from a reputable camera manufacturer. The reason for the overexposure was simple: the camera, perceiving a lack of ambient light, opened the aperture completely, ignored any auto-focus given distance and the fact that the flash was on. The built-in flash would have been controlled nicely using the camera's TTL metering, except an external flash was used, synchronized optically, that is, the external flash was a slave to the small built-in flash. Without a shut-off signal, nor ability to tone down the flash, the external flash fired fully, with the given results. With limited camera abilities there was little the photographer could do with his setup.

Backscatter and its solution

The built-in flash of digital cameras is largely a waste of time for underwater photography (with some exceptions!) The flash is too close to the lens, too much in-line with it, that backscatter is virtually unavoidable. Furthermore, the flash is not meant to light up the light-absorbing, dark and dense water medium divers take it into. Illustration 1 shows a very nice example of backscatter. It is for these reason that most underwater cameras have powerful flashes which at rest are a distance away from the lens, and are usually connected by a cable to the camera, allowing the photographer to move the light at will. Illustration 2 shows the same scene as in illustration 1, taken a few seconds later, but using an external flash. Notice the complete lack of backscatter. In order to achieve the same results the user of a digital underwater camera has no choice but to follow the same route. Unfortunately, this is where the real problems set in.

Slave flashes on digital cameras

Except for the top-of-the-range digital cameras, virtually none of the more reasonably priced cameras have a flash hot shoe, or if they do, no housing is available for that camera model. Thus to achieve a remote flash there is only one way to fire that flash: optically. This method has been in use for a long time, and such flashes are called slave flashes, as they rely on another flash to fire first. Several companies, notably Sea&Sea and Ikelite offer either slave flashes or slave flash triggers. Even the Nikonos SB105 (as well as the old SB102) can be used as slave flashes, having the optical triggers built in. The other companies offer different solutions, with either a trigger on a cable, on the flash, sometimes with an optic-fibre strand from the camera flash to the flash, to ensure it triggers. These solutions work, except for one small, and rather annoying feature, of most compact and digital cameras: pre-flash.

The pre-flash dilemma

My Canon EX380 already offered this "feature", where the flash would pre-flash, in order to evaluate the exposure for better results. It probably works very well - calculate how much light is required, and allow the TTL to fine-tune the result. There is just one small problem: a slave flash will trigger on the pre-flash first, and not be charged for the subsequent flash, and hence not light up anything. This can be overcome - the companies mentioned earlier have dedicated digital flashes which can be set to ignore pre-flashes (and some cameras have two or more!), while other companies have created just slave flash triggers which can be set in a similar way, and even a hobbyist offers a waterproof, just-plug-in-your-flash version. After looking at these solutions I concluded that they were not ideal: remember the camera's behavior: open the aperture, fire the flash. The results are probably acceptable, but for macro work depth of field is essential, and cannot be compromised upon. Suppressing the pre-flash is thus not the ultimate solution, although this does not mean it might work for you.

Working with slave flashes

My digital camera, a Canon PowerShot A70, thankfully has a way of switching the pre-flash off: put it in full manual mode. This may sound like asking for trouble, but remember that automatic modes are rather pointless the moment you start using flashes (slave or not.) In full automatic mode the lack of ambient light will cause the camera to choose a wide-open aperture and a slow synchronization speed (shutter speed) for the flash; in aperture priority the shutter speed would remain slow, irrespective of the aperture set, in shutter priority the aperture would be wide open, irrespective of the shutter speed set (well, mostly irrespective, but do not forget shutter speed must often be within a certain range for it to work as flash synchronization speed.) So, you might as well be in manual mode! The beauty of digital cameras (and this is why the solutions offered would not really work well in equivalent film happy-snappy cameras in housings) is that you can see your results immediately, and adjust accordingly immediately. So, with a flash which fires at fixed strengths (remember, there is no more TTL!), you slowly over time will build up a mental exposure table which will give you the results you want and expect.

My solution Flash trigger

Rather than buy a digital flash, or buy a slave trigger, with or without pre-flash suppression, and with or without waterproofing, I decided to build my own. Having discarded the pre-flash suppression as a requirement, all I needed was a simple slave trigger. The circuitry was designed by a colleague, Charl Cillie, and I tinkered with it until it worked correctly. The circuit is shown in illustration 3. Illustration 4 shows the Nikonos flash plug connections. Instead of using a battery I used the flash ready light as a source of power. The cost of building is minimal.

In order to take it under water, I had to waterproof it, that is, build a housing. I used a piece of perspex tube with an appropriate inner diameter (12mm) to fit the plug on the Nikonos flashes' cable. The plug, which has an O-ring, plugs into a socket which I took from a damaged Nikonos camera, but would not be hard to make oneself, while the other end of the tube is sealed using a blanking plug from a Nikonos camera, which can be used to seal the socket when no flash is used. As I could not (I tried) get a thread cut into the tube, I just rely on water pressure to keep both plug and blanking plug in, but while the plug goes in far enough, the blanking plug does not and could be dislodged much too easy. To overcome that scrounging around the house revealed some plastic tubes which fit exactly - these are the white caps shown in the component and assembled view in illustrations 5 and 6. The housing for the Canon A70 now has an add-on for a removable diffuser for the built-in flash - this I conveniently used for my trigger unit: the black plate of plastic shown in illustrations 5 and 6 slots into that add-on in place of the diffuser, at the same time preventing the built-in flash from reaching anywhere but the trigger.

The whole camera and flash assembly are shown in illustrations 7 and 8, where the rear view shows the diffuser hanging down. I keep the diffuser with me, as for macro work the build-in flash is often very useful, and thus allows me to control my light even more by allowing me to use the internal flash where required. On the base-plate I tend to mount a torch (now with diffuser too, to give a more even light), which I use to see things, and the auto focus uses to focus.

The results

Illustration 2 was taken using the setup shown in illustrations 7 and 8. The Nikonos flash was set to � power, was without diffuser (the white cap shown on the Nikonos flash in illustration 7), the camera aperture to f8, the shutter speed to 1/60 and the ASA setting to 100. This means that there is still sufficient play to get more light for more distant objects, yet also enough for macro work.

To save batteries on the camera its flash was set to the lowest setting, as little flash is required to trigger the external flash.

Improvements

The following variables are available:

  • Nikonos flash power: full, �, 1/16.
  • Nikonos flash diffuser.
  • Camera aperture from f2.8 to f8.
  • Camera shutter speed.
  • White balance.

Ideally the aperture should be kept as closed as possible to improve depth of field, although auto-focus allows for much easier focusing and hence less errors. The Canon A70 uses either a 5-point auto focus system, or a manual override to the central focusing point - the latter is very useful in many instances.

Changing the shutter speed has limited effect when using flash, but can be used for special effects, specifically, motion. While the camera manual does not specify, I would assume the flash simulates "front curtain synchronization", that is, it fires on starting the exposure. A more natural effect can be achieved using "rear curtain synchronization", where the flash fires only just before ending the exposure - the difference is that the object in motion is either frozen at the start or the end, where the latter is the better one.

White balance can be used to change the colour of the image. For example, the, Nikonos flash diffuser, being opaque white, creates a yellowish image. This can be compensated by using the "tungsten" (light bulb) setting.

The strobe arm shown in illustrations 7 and 8 is the standard Nikonos strobe arm. It has served me very well with the Nikonos V so far, but for the digital camera it is somewhat impractical. With the Nikonos V the photographer decides in advance what type of photos to take, be it macro or wide-angle. Depending on that decision the strobe is placed in different locations: for macro straight above the camera, for wide-angle as shown in illustrations 7 and 8. With the digital camera and its zoom lens, with very capable macro facilities, this is rather limiting. Here there is a need to move the flash quickly, from a macro setting to a wide-angle setting from photo to photo. Furthermore, it must be also be possible to move the flash away when using the built-in flash, which I find very handy in tight macro situations. To solve this problem I adopted Richard Steventon's solution, that of a Loc-Line arm. The idea of using Loc-Line is not new, and it has been used extensively especially with underwater video, but the cost of these arms is generally rather high. Not until Richard explained the origin of Loc-Line (being engineering workshops) and showed how well it worked (see illustration 10) did I try it as well.

The Loc-Line arm is not strong enough to carry the flash on land - illustration 9 has the arm carefully placed in location, with the flash lightened of batteries, but the positive buoyancy of the torch underwater means that there is no problem. One must take care going into the water and coming out again, as the whole unit "collapses" easily, and requires careful carrying.

Apart from the quick relocation advantage, the arm now also allows fine-tuning of the quantity of light, by moving the flash towards or away from the subject.

An alternative idea: video light A video light assembly

Richard Steventon decided to go a different route. His camera, a Canon A20, allows for less control than my A70, and especially has no means of over-riding the double flash. Furthermore, he decided that rather than buying an external strobe he would rather use something which he could use otherwise. The result he came up with is essentially a video light. Technically it is a cannister light: a small light head holds a 12 volt down-lighter, fed power from a rechargeable battery in a cannister, which is carried on the body. The idea here is that if he could not add light on demand, just keep the light on! The light is mounted on a flexible arm. Illustration 10 shows how the assembled unit works. Shown is also the close-up lens and framer he needs to use due to somewhat lacking macro facilities (compared to the A70). Illustration 11 and 12 show Richard in action. On illustration 11 the cannister housing the battery for the light can be seen - it is the blue cylinder on his belt. A long flexible cable ensures that his movements are not restricted, and the ability to quickly detach the light head allows him to peek into caves and crevasses. The setup is thus very versatile, and works very well.

The results

First trials using the light shows that the colour was much warmer than using a flash, even though a compensating filter was used. Illustrations 13 and 14 show the same fish, where illustration 13 uses the camera (Canon A70) flash, while illustration 14 uses the light. Both images are perfectly sharp and well lit.

Improvements

The warmth in the colour can be compensated using white-balance, allowing for a more natural look - this was proven in subsequent dives. The advantage of using a video light is that there is always enough light for the auto focus to focus easily, and it helps the photographer to see the subject at any time, night or day! The ability to light up larger areas, especially during the day, is somewhat limited, especially as anything more distant would result in colour loss.



Alternative Solution: using the Sea&Sea YS-40A strobe for the MX10 camera

The MX10 camera from Sea&Sea was its bottom of the range model at the time it was released. The camera came coupled with a small strobe, the YS-40, which optically synchronized with the camera. For that to work the flash had to be attached to the camera, and remain attached. The synchronization is in infra-red, but normal flashes have enough infra-red in them to trigger the flash, and I have used this small flash successfully as a slave flash both above and below water. There was one downside of the original camera and flash and that was that the flash did not have any power settings, and the camera has very limited settings; TTL was not available. This meant that using the camera was often more a miss than a hit operation. Sea&Sea must have realized this, as they brought out an improved flash, the YS40A. The difference was a modified flash "lens", and pseudo-TTL. Illustration 15 shows the camera and its flash. The pseudo-TTL is not true TTL (where TTL stands for "through the lens") in that the flash, rather than the camera, picks up the amount of light reflected and switches off. The eye of the sensor can be seen on the flash in illustration 15 - it is the small black round eye below the flash lens on the arm. The idea is simple: put the flash on "auto" mode, put the camera, given the film speed (and the MX10 only has the option of ASA 100 and 400), on a certain aperture, and the flash will take care of the rest. I have seen the results, and it works quite well. The flash, being a small unit, is of course not the most powerful, but it is sufficient for normal shots and great for macro. With macro the chance of over-exposing is much greater, so the pseudo-TTL becomes a very valuable feature indeed!

The idea now is to use that little feature for our digital cameras. The biggest problem is that the flash does not have any method of attaching to anything but the camera it is meant to attach to. This makes it hard to put it on the baseplate, with or without a further extension arm. Furthermore, any attachment must keep the synchronization eye on the base of the flash arm within line of sight of the cameras flash.

To overcome the possibility of not triggering the flash it is certainly possible to create a cable to trigger it - the same flash trigger sensor shown earlier (illustrations 5 and 6) can be used to fire an infra-red LED (light emitting diode) directly onto the flashes' synchronization eye. This requires some work however, as now two housings are required: the slave trigger sensor and the actual trigger, coupled by a cable. A cable is handy, as it means the photographer does not need to worry about where the flash is. Of course, if you keep that in mind it is possible to optically synchronize for quite some distance - on land I managed some 5 meters or more, although admittedly I fired my main flash straight at the YS40A. A different solution is used by several camera manufacturers, namely optical synchronization using a fibre-optic cable.

To test it with the digital camera I created a silver flash reflector out of an old CD, and placed it in the Canon A70's housings flash diffuser slot. The slave flash fired in virtually all hand- held positions, as long as the flash is to the left of the camera. Of course, optically synchronization, be it line of sight or via a fibre-optic cable only works with cameras where the double-flash can be suppressed, such as in the A70's manual mode. Illustrations 16 and 17 show how well it can work. Please excuse the land shots! The setting on the camera was 1/125 shutter speed, aperture f8 and camera flash on its minimal setting, all on manual mode. On illustration 16 I switched the pseudo-TTL off - as can be seen the white of the stryrofoam is over-exposed, whereas on illustration 17 it is not.

To test it underwater I simply attached the strobe to my strobe arm, relying only on optical synchronization. The latter is a problem: with the synchronization eye being on the other end as the flash head the unit is a bit cumbersome to use, and especially hard to get into tight spaces or use really well for macro. Numerous photos were really underexposed simply because the YS40A did not fire! The lack of power is another problem, of course. That said, illustration 18 shows what can be done: the image is well exposed and the colours look right. It did help that the fish (a speckled klipfish, Clinus venustris) posed happily, and that this is the best shot of about 30 taken. This is where the digital camera comes into its own - with a normal 35mm film camera 36 is the maximum, and the average underwater photographer would rarely do this!

Digital Strobes

Several manufacturers, notable Ikelite and Sea&Sea, have created what they call "digital strobes". Initially I thought these were just slave strobes which suppressed the pre-flash, but they are "brighter" than that, if you can excuse the pun. Instead, they mimic the pre-flash. The pre-flash in digital cameras is used for more than just setting the required flash power, but also, and possibly more important, the white balance. By mimicking the pre-flash, the camera will receive the same information from the subject as it would using its normal flash, and hence get the right colour balance.

I have not had the pleasure of using such a flash, but I can imagine it will work extremely well. One thing to remember though is to not use the flash in the normal program mode (the "automatic" or "dummy" mode), as this will give you too little depth of field especially for macro: remember that the camera sets aperture according to ambient light, not possible light with flash! In this case, given that the camera will sync at specific set speeds with a flash, use the aperture priority mode, and for macro close up the aperture as much as possible - with a bit of luck the flash will come with a chart as that provided with the Nikonos flashes.

Previous Articles Index