Flashes, Continuous Lights, And Flash Triggering Modes
It has been discussed and debated by many fellow extreme macro photographers whether to use flash or continuous light for extreme macro photography. On one side, people think that flash is better because flash has shorter duration and can freeze an image. On the other side, particularly at extremely high magnification, people find it continuous light is better and have shown so with their excellent work, though it seems contradictory to the belief that any vibration during exposure can ruin it.
So the debate of a seemingly intuitive thought of flash versus some excellent works done using continuous light goes on. And very often, people who use continuous lights claim that they can not do the same excellent work with flash and this is often verified by many others, particularly at high magnification.
When I challenged myself to do some high magnification stack, I tried to stack some butterfly wing scales at 50x. Having removed all other factors, such as putting all equipment on a large and heavy granite (25kg), locking everything down, mirror lockup, timer triggering, etc, etc, almost every trick that I heard of, I still was not able to get good results, as matter of fact, it is not even worth showing these results. I even doubted the cheap objective I used (Nikon 50x 0.5NA, bought for 140USD, rather cheap for a 50x objective). But another factor is that I used flashes as light source. Feeling desperate, I bought 3 Ikea Janjso LED lights and started stacking with long exposures, often 2 seconds exposure. To my surprise, even though the results were not as amazing as work done by other experts, they were far better than using flashes. Here is an examples:
So what is going on? This is really counter intuitive, to say the least!
From the experience of challenging myself to do high magnification stack, I have learned that the tiniest vibration can easily ruin your stack. For example, at 50x (or even at 30x) even if I have a heavy 25kg granite block as base, walking around the setup will cause enough vibrations that you can see on live view screen. If you put your finger on the camera, not pushing it, just touching it, you will see vibration on live view screen. Breeze from air conditioner will cause butterfly wing to move and you can see it on live view screen. And if you turn your stereo on, you will see your butterfly wing dances to the tune on live view screen.
So, to get a good stack at high magnification, keeping vibration down is the key. Then, what is the difference between using flashes and continuous lights? Well, when using flashes, there are a lot going on inside the camera at the moment exposure is done! This is particularly true when shutter speed is set to just enough to synchronize with flash, as one would do in normal photography, say 1/160 or 1/200 of a second. So what is going one here? Here is the sequence of events happening, assuming mirror is locked up, when an image is taken:
Shutter start to move to open -> fully open -> flash triggered -> shutter move and closes.
However, shutter is a mechanical device, the mechanism to move the shutter will cause enough vibration, particularly, shutter assembly is inside the camera and so close to the sensor. Furthermore, most modern cameras use two shutter blades and they move in sync to achieve high shutter speed.
Here is a video showing how DSLR shutter mechanism works in slow motion by The Slow Mo Guys.
But why do continuous lights work? You still have shutter blades moving inside the camera, they still cause vibration, what gives? The answer is rather simple: when using continuous lights, very often exposure time is much longer than using flash, typically, it is one or two second exposure for continuous lights. Why does it help? Even though shutter blades and the mechanism of moving them causes vibration, the vibration does not last very long, probably in milliseconds (this is pure guess from some experiments done below
. Say the vibration lasts 100ms and exposure time is one second, the vibration only contributes about 1/10 of the exposure time, thus reducing the effect of vibration. On the other hand, when using flashes (and normal first curtain sync), 100% of the exposure is done during the vibration period, even if flash duration is short, at high magnification, it is not short enough.
First (Front) Curtain vs Second (Rear) Curtain Flash Modes
In above analysis, I mentioned the term "First Curtain" or "Front Curtain" if you are Nikon user. What are they? First (or front if you are Nikon user) curtain is a flash synchronization mode where flash is triggered when the first shutter blade is fully open where image is then exposed and captured. Second (Rear if you are Nikon user) curtain is another flash synchronization mode where first shutter blade is fully open and the second shutter blade is about to close.
Why does it matter? Well, as first curtain (front) sync mode implies, the flash is fired when the first shutter blade is fully open. However, at this moment, the whole shutter assembly has NOT settled yet, things are still vibrating, so, it is NOT the good moment to expose an image. Shutter speed is irrelevant here because the flash is always triggered at the very beginning where vibration is most damaging.
OK, so what about second (rear) curtain? From the definition of it, it seems very feasible that if we slow down the shutter speed, let the flash being triggered at the end of exposure (vs at beginning for first curtain) so everything is settled already, we might be able to avoid, or at least reduce, the vibration.
For example, lets set shutter speed to one second, when we take a picture, first shutter blades opens up and then settles down and stabilizes in, say 100ms, then the flash will get triggered at the end of the one second. This sounds great, though, according to definition, but mostly it depends on actual implementation of it, the flash is triggered when the second shutter is about to close. So does the flash get triggered when the second blades has moved already or remaining still? This is probably up to the camera model and manufacturer. But this mode does help significantly.
Here is another slow motion video showing how DSLR shutter works (credit CameraTest), notice how the first curtain opens and then the second curtain follows.
Electronic First Curtain Shutter
Many cameras has a special mode called electronic first curtain shutter mode, EFCS. In this mode, the mirror is already lifted up and the exposure is done without any mechanical movement. When exposure is done, the second curtain comes down to finish the exposure. In this mode, image quality improves significantly.
There is one catch for Canon users, that is, it is almost impossible to set flash trigger mode to second curtain, Canon simply does not have that function for most of its cameras unless a TTL enabled flash is used. So, essentially, it is impossible for Canon user to do this.
Another catch for most cameras when true electronic shutter is used is that the flash is disabled. This is where the new feature of the new controller comes in -- it fires flash independent of camera.
Words, words, and words, all of above are words :-). Lets see if we can do some experiments to show it all.
First Experiment - lets take a look at what is going on by using an vibration detection App on iPhone -- Physics ToolBox. To do this experiment, an iPhone is mounted on top of a Canon 550D in its hotshoe. Here is the setup
Here are some images captured while triggering the camera with Physics Toolbox app running. Some explanation is needed: the red line measure vertical vibration, the blue line measures back and forth vibration, and finally the green line measures left and right vibration. Important note, this is true when iPhone is placed on top of the Canon 550D like the setup. Camera is set to mirror lockup mode and 2 seconds timer is enabled to allow mirror to settle and trigger camera again required by mirror lockup.
The first one is when camera, Canon 550D is set to mirror lockup mode without live view. Notice that, when camera is triggered to take a picture, mirror is lifted and vibration is captured by the app, after 2 seconds, timer triggers the second time and first curtain is open and exposure starts and as expected, vibration is captured again. After one second exposure, the second curtain starts to close and, again, cause vibrations.
The second one is when the Canon 550D is set into EFCS (electronic first curtain shutter) mode, ie, the camera is placed into live view mode. Note, when timer light is off (not seen here), it means exposure starts already. But here, we do not see any vibration until the second curtain starts to close. Being in live view mode, the camera must lift the mirror and second curtain up again and that is where we see a second vibration. But these two vibrations do NOT cause any image blur because they occurred AFTER exposure. From the graphs above and below, we can see that vibration does not last very long, particularly looking at the above picture, it seems the vibration lasts about 1/10th of the total exposure, which is 1 seconds. This means the vibration is about 100ms.
What does this mean? Actually, it means if we can trigger flash in the middle of exposure, after shutter vibration has settled, we could get pretty good results, even if we do NOT use EFCS.
Now first stacking experiment: heavily repaired Canon 550D camera, two studio strobes (long flash duration), Olympus MSPlan ULWD 20x on Pentax 105mm as tube lens. The image on the left was done with "normal" flash triggering. On Canon, that means the strobes were triggered at the beginning of exposure, ie, first curtain. The image on the right was done using the new controller which can fire flashes on its own. The strobes were triggered in the middle (towards end) of 1.3 second exposure. The difference is huge.
Second stacking experiment: some might argue that the camera is heavily repaired and does not represent a true "normal" use case. Fair enough! I just happen to have a brand new Canon 6D and see what happens. Camera is Canon 6D, Mitutoyo 20x on a Pentax 105mm as tube lens. The left image was stacked from 121 images done with "normal" flash operation while the right image was stacked from 121 images done with "mid-curtain" flash (Open it in full size to see it clearly)