Flash duration – what's it all about?
This blog has been prompted by a forum question on a photo forum, Talk Photography. You’ll see that the question was about freezing subject movement, in this case the fast-moving hand of a child. There was blur in the photo because the flash duration of the budget studio flash was too long for this particular subject.
As usual, there were some answers that made sense and others that were totally wrong, plus the usual opinions from people who don’t actually understand the basics – so let’s start by clarifying the basics.
When you use flash indoors, any subject freezing action is created by the flash itself, and the shorter the flash duration, the greater the ability it has to freeze any movement of the subject. The shutter speed makes no difference, because the sole function of the shutter is to be fully open at the time that the flash fires.
In a totally dark room, it makes absolutely no difference what the shutter speed is, as long as the shutter is fully open when the flash fires, therefore the shutter speed has no effect on the finished result.
If there is ambient light in the studio, for example from daylight entering the room, from modelling lamps or from room lighting, a long shutter speed may allow some light pollution to effect the shot. Because of this, we normally shoot at around 1/125th of a second, but we only do this to prevent light pollution, because slow shutter speeds may allow the ambient light to contribute to the exposure and cause light pollution. All of the answers in that forum thread that advised the use of a faster shutter speed are just wrong.
If there really is a lot of ambient light present then a fast shutter speed will help to reduce its effect, but as most modern cameras have a focal plane shutter, consisting of two moving blinds, there’s a limit to how fast that shutter speed can be – go too fast and the second shutter curtain will have started to close and part of that second curtain will appear in the shot – this short video explains the shutter action.
Typically, the fastest shutter speed that you can use will be around 1/200th to 1/250th second, depending on the camera used, but this can be reduced if you use a radio trigger that has an inbuilt delay in its system, as some of them do. This isn’t really something we need to cover in this blog, which is just about flash duration, but you need to be aware that because of this it isn’t always possible to achieve the maximum theoretical synch speed for your camera.
Before we concern ourselves with flash duration, we need to be aware that there are two very different technologies available, conventional flash heads and IGBT.
Conventional flash heads can be adjusted to output a range of power, typically from full power down to 1/32nd power, (although this varies). When the flash is fired, the power quickly reaches its peak, then dies away fairly slowly. When measured with an oscilloscope, it looks a bit like the graph on the left, except that it isn’t coloured blue through to red, I’ve done that bit myself to show how the colour of the light changes as the light tails off in intensity, the flash starts off a bit blue and ends up a bit red, and the average should be white, which is what you want it to be. I’ve exaggerated the colours greatly for the sake of effect.
With conventional flash heads, which is nearly all of the studio flash heads, if you dial down the power then less power is passed through the flash tube, which produces less light. Turn it down to 50% and the energy peak is at 50%, as you’d expect, but because the most common way of reducing the power is to reduce the voltage, the burn process takes longer too, and this results in a longer flash duration. Also, the flash doesn’t start off quite as blue, so the average ends up more red, which you don’t want.
If you want to know just how typical these graphs are of actual flash heads, they aren’t typical, or accurate, at all, they are just representations of how it works. Really cheap flash heads tend to have much longer flash durations and much greater colour shifts than high end ones, and some high end ones adjust their power output by switching out capacitors as well as just reducing the operating voltage, so there is a lot of variation out there.
And the other technology is IGBT. IGBT is the abbreviation of Insulated-gate bipolar transistor. With IGBT the flash ALWAYS fires at full power, and it theory it always fires at full voltage too. When you fire an IGBT flash at full power the graph looks similar to the one above, and the flash energy decays in much the same way as with a conventional studio flash head. But when you set a lower than full power setting, it behaves very differently, as shown in the graph on the right. The first bit of the graph is the same but then the flash is cut off, or quenched, and instead of gradually dying away, the flash comes to an abrupt stop. Quenching the flash in this way does two things – the obvious one is to reduce the amount of light reaching the subject, and there is also the less obvious benefit that the flash burns for less time, so the flash duration is shorter. And it can be dramatically shorter, every time you reduce the power setting by half, the flash duration divides by two as well, giving the potential for very short flash durations. This graph shows a typical picture of the way that an IGBT flash behaves at various power settings. As you can see, the flash always fires at full power and is simply cut short at reduced power settings. This has an extra advantage, because at reduced power settings most of the stored energy is left in the capacitors, so recycling speed is dramatically improved too.
IGBT is obviously much better if you need fast flash – why isn’t it universal?
IGBT technology is very cheap and easy when used with hotshoe flashguns, which is why all modern flashguns use it. But it’s complex and expensive when used with powerful studio flash heads, and because of this very few manufacturers, notably Alien Bees with their Einstein model and Lencarta with our SuperFast models, have been able to overcome the technical problems and produce affordable IGBT studio flash heads. Most manufacturers stick with the conventional technology, it’s cheap and easy and is good enough for most studio flash work.
There are a few makers of conventional flash heads, right up at the top end of the price range, that have products with pretty short flash durations. They achieve their high performance through good engineering design, by having physically small flash tubes, by operating at very high voltages and so on, all this adds to the production cost and also places the flash head under increased stress, but they do tend to perform better in terms of flash duration than the cheaper flash heads, although they can’t achieve anywhere near the speed of the better IGBT flash heads, nor can they recycle anywhere near as quickly as IGBT flash heads.
Just how fast does the flash have to be to freeze subject movement?
There’s no answer to that because it depends on
The speed of the movement
The direction of travel
The degree of magnification
Speed of movement is a bit obvious, but the direction of travel and the degree of magnification are just as important. A F1 racing car, doing 200 mph some distance away and moving more or less towards the camera, is easier to freeze than an ant, photographed at twice life size, moving across the frame. As soon as you magnify the size of the subject, you magnify the movement blur too. And when the subject is crossing in front of the camera that movement becomes much harder to stop than when it’s moving towards or away from the camera.
And some subjects just can’t be frozen with even the shortest photographic flashes. We’ve all seen photos of bullets apparently being shot into things, with everything pin sharp, but in reality it just can’t be done with even the fastest IGBT flash. The reason for this is that bullets move extremely fast, for example the SLOWEST real bullet is a .22 rimfire, and even the slowest of these move at around 1000 feet (304,800 millimetres) per second. In a life-size image (and most of these shots are bigger than that) that’s a bullet movement of 10.16mm during a flash duration of just 1/30,000th second! The way that these shots are actually produced is either to use a piece of very specialised lab equipment known as a spark generator (similar in principle to a spark plug used in a petrol engine) and also known as an air gap flash – which produces an incredibly short burst of light – or by faking. Faking can of course be carried out on computer, but another way of doing it is simply to remove the propellant from the cartridge, leaving just the primer to fire the bullet, at very low speed. That’s why these shots often show a bullet passing through something like an apple – in reality, the apple would simply explode when hit with a real bullet fired at normal velocity.
But shots such as the ones on this page can easily be done with IGBT flash heads. For these shots, we used SuperFast flash heads set at low power, the flash duration will have been around 1/20,000th second (t.1) simply because we wanted to use the camera in continuous shooting mode at 8 frames per second and so needed to set the lights to low power to get ultra fast recycling, but that evil mix of milk, yoghurt and food colouring wasn’t really moving all that fast, and we could probably have frozen the action at something like 1/3,000th second. The hair shot could probably have been done at about 1/3,000th too, but the shots of the balloon bursting and the scarf being flicked really did need a very short flash to freeze it well – so, there are only a few types of subject that really do stretch the capability of the SuperFast at 1/20,000th second. If you do want to photograph movement too fast even for the SuperFast, try using an ordinary hotshoe flashgun at minimum power – they may be even faster, although of course the amount of power they are able to produce is extremely low.
I mentioned the colour variation as the flash energy degrades – what’s that about?
Well, with conventional flash head technology, the flash very quickly becomes blue and then gradually tails off towards the red end of the visible spectrum as the power dies away. Generally, the mix of cold (blue) and warm (red) colours mix to produce a fairly good white. When the power is turned down however, it starts off less blue than it should, and ends up much more red than it should, at least with the cheaper flash heads, and the final result ends up looking warmer than it should. The colour often also varies dramatically from shot to shot. A variation throughout the power range of no more than 300K is usually considered acceptable for portrait use; all of the good makes achieve this, but some of the cheaper makes can vary by as much as 1000K. Cheap makes often claim on their website to be colour consistent within something like 100K, but that’s just not true.
As you know, with IGBT technology the flash always fires at full power (so has the same bluish colour) and although the flash degrades in colour as well as intensity, just as it does with conventional flash heads, when the power is reduced the tail end of the flash is cut off. It isn’t just the tail end of the power that’s cut off, it’s the red-ish colour too, which if left unchecked would result in the light becoming more and more blue as the power setting is reduced. That does in fact happen with many hotshoe flash guns, and with certain IGBT flash heads that I won’t name. But the better makes, including Canon hotshoe flash guns, the Lencarta SuperFast and the Alienbees Einstein, all have a clever compensating mechanism built in that in effect reduces the input voltage as the power is reduced, this keeps the colour temperature stable and as a result the colour temperature consistency of the best makes is actually better than with conventional flash heads. It isn’t just acceptable, it’s incredibly good!
We’ve been talking about flash durations. But what does this mean?
Let’s look at the graph again. When the flash is fired the tube output rises to maximum brightness pretty quickly, as the graph shows. This is followed by an decline in light as the capacitors are discharged to zero. The standard engineering term for stating flash duration is “t.5”. This describes the time it takes for 50% of the total flashpower to be discharged. Whenever the simple designation “Flash Duration” is used it’s usually safe to assume that the figure is quoted as t.5. The reason for this is that this is the usual way that manufacturers describe the duration of their flash heads, it also produces a figure that seems higher than it really is, and the high figure suits the sellers.
My suspicion is that many manufacturers, and pretty much all resellers, don’t test flash durations at all. They just quote figures that sound good but which are very unlikely to be accurate. If they did test them, then surely they would explain how the tests were carried out and show the results instead of just quoting unlikely figures…
The real problem is that the t.5 spec doesn’t predict the actual motion freezing capability of a flash. As I explained earlier there is a much longer trailing edge that carries on discharging the remaining 50% of the light. This causes a lot more motion blur than the t.5 spec implies. In order to more closely match the flash duration to an equivalent shutter speed, the term “t.1” was introduced by the photo industry. t.1 specifies the time it takes for 90% of the total flash to be discharged. The remaining 10% of flash energy is really pretty insignificant and can be safely ignored. A lot of people believe that the t.5 time is about 3x that of the t.1 time, so if you divide the stated t5 time by 3 you’ll end up with the t.1 time. For example, if the t.5 time is 1/1500th then the t.1 time is going to be 1/500th.
But it doesn’t work out like that. It more or less does with the cheapest flash heads, operating with inferior capacitors at low voltage, but the better ones are closer to a factor of 2, so a t.5 time of 1/1500th is likely to have a t.1 time of around 1/750th.
But even the t.1 time doesn’t accurately predict action-stopping potential. To find out how fast the flash really is, compared to a given shutter speed when used without flash, we need to carry out tests.
Basically there are two main testing methods, the scientific one and the practical one:)
The scientific method involves using an oscilloscope to actually measure the duration of the flash. This is a meaningful method to anyone who is involved with either electronic engineering or physics but it isn’t perfect because
- Many people find it hard to understand oscilloscope trace readings
- Many people find it hard to interpret the actual info and relate it to action-stopping potential
- Most people don’t have access to an oscilloscope, so they can’t check the accuracy of seller’s statements for themselves – and their use requires training.
- Oscilloscopes can’t measure really fast flash durations accurately.
So, here at Lencarta we have moved over to a method that we believe to be far more useful – we photograph a revolving disk, it’s a CD, fitted to the spindle of a fan that has had the fan blades removed.. This method produces more meaningful results because people can see just how well the flash freezes the movement of the disk – and they can, if they wish, check it themselves – fans cost a lot less and are far easier to use as a testing tool than an oscilloscope. Of course, the moving fan blades created a lot of air resistance and slowed the fan down; with the blades removed, the fan is moving at incredible speed, far to fast to test the action-stopping potential of many flash heads, so a potentiometer has to be introduced to the circuit to slow the fan down. This is of course adjustable, and we can adjust the speed to suit pretty well all flash heads, whether they have conventional or IGBT technology.
What we do is to photograph the disk using continuous light, at various shutter speeds. This gives us a benchmark showing just how sharp the figures are at each shutter speed, say from 1/125th second through to the fastest shutter speed available on that camera, say 1/8000th. As continuous lighting is pretty dim compared to flash, we need to juggle camera ISO and lens apertures to get correct exposure at the various shutter speeds, and this juggling does affect the appearance of the test photos, but it’s good enough for the purpose even at very high ISO settings.
We then photograph that revolving disk again, but this time using our flash head or flashgun at the only light source, the camera shutter is of course now at a normal sync speed, say 1/125th second. We take a shot at various power settings to see what effect the power setting has on the flash duration, and we then compare the results with our benchmark images taken using continuous lighting. It isn’t very scientific but it is extremely useful, because if a shot taken with flash looks the same in terms of blur as a shot taken at (say) 1/1000th second using continuous lighting then we can safely say that the flash duration, in terms of shutter speed equivalent, is 1/1000th second – we can ignore the theoretical t.1 and t.5 flash duration times.
There is in fact a third method of testing flash durations. It’s crude, the results aren’t all that accurate and for people whose camera has a focal plane shutter and who don’t have a flash meter, it only works with fairly slow conventional flash heads, but it’s a method that anyone can do at home.
It’s called the gate method, and involves taking a series of shots using a flash head with the modelling lamp turned off, and in darkness.
Take the first shot at say 1/60th second, then 1 notch up (usually 1/90th) then repeat at say 1/125th, 1/160th, 1/200th etc. Then compare the exposure of those shots on your computer. If, for example, the shot at 1/250th second is measurably darker than the ones at the lower speeds then you know that the flash lasted for longer than 1/250th second because not all of the light was captured at that shutter speed.
You can then repeat that test at all possible power settings, and you will end up knowing the flash duration of your studio flash heads at all power settings. Of course, this testing protocol is limited by the max synch speed of your camera shutter, so it isn’t any good for testing IGBT flashes or even for testing fast conventional flash heads if your camera is fitted with a focal plane shutter. But if you have a flash meter, use that instead, you will then be able to test your flash at all of the shutter speed options on your meter.