My progress in developing my water drop collision equipment
My original interest in high speed flash photography stemmed from trying to capture foxes in my garden at night. This was just before Christmas 2011 when we had been watching foxes foraging for food in the garden that my wife had left out for them.
For this I needed some kind of trigger to release the shutter on my camera. After some research with a friend I found the Smatrig made by Luk in Germany
Reading on his website I notice some information on using high speed flash to capture water drops splashing into a bowl of water. This really captured my interest more so than the idea of photographing the foxes so I decided to have a go.
I read a few articles on the subject and decided to build my own system for doing it, using my computer knowledge for controlling the system, my diy skills to build it, and using the Smatrig to trigger the camera
Before I embarked on this project I had no idea how it could be done and was not aware of any systems that were available for purchase for doing water drop photography. Nor had I done any electronics since building crystal sets (radios for the younger readers) when I was 10 years old. It was therefore going to be quite a learning curve for me
I need some way of controlling the system and I found that an Arduino mini computer would do this.
It is attached to a pc by a usb cable which is used to download a script, written on the pc to program it.
Now things were getting a bit tricky as I had no idea on how to write a computer program, but with the help of the internet I managed to write a simple sketch for the Arduino. When the sketch has been written the program verifies it and then uploads it via the usb cable to the Arduino. The Arduino then runs the program over and over again until the power is switched off. It can be left powered via the usb cable or using a 5 volt supply. The Arduino is very simple but can be made to do all manner of things. What I wanted it to do was to switch the solenoid valves on and off at the right time to control the timing and the size of the water drops. The Arduino provides output to the digital pins which in turn worked the relays on the relay shield fitted on top of it, which in turn operated the solenoid valves .
Next I wanted some solenoid valves. A search on e-bay found some in Hong Kong at about £4.50 each. When these arrived I mounted them above my work bench and made some nozzles for them to provide the drops.
The water bottles are mounted above the nozzles and the rate of flow is adjusted by the height of the water level in the bottles above the nozzles (the head)
It is important to understand that as the water level in the bottle supplying the water to valves diminishes, so does the head, and so the flow diminishes. This can be very problematic when dealing with precise sized drops of water. To overcome this you have to use a mariotte siphon. You know how a siphon works, but as the water level in the container drops so does the head and so the flow. The mariotte syphone solves this problem by having the water in a sealed container, with a breather tube from the outside to the bottom of the bottle. As the water is drawn out of the container the air is pulled into the bottle via the breather. The air pressure at the bottom of this breather tube is at atmospheric pressure as is the water at the outlet of the nozzle, so the head remains constant as the water is drawn off. The head being the distance from the bottom of the breather to the tip of the nozzle. The bottles are of plastic, and two holes are drilled in the screw top through which two 5mm acrylic tubes are passed and cemented in place. The bottoms of the tubes are about 5mm off the bottom of the bottle.
If the head is too great the flow increases. This presents problems when using small nozzles because the shortest time for the valves to actuate is about 10 milliseconds, and with a high flow rate this might be too long for the creation of a single drop. Reducing the head to about 200mm works for my system with water, but with other liquids you may have to adjust it.
I have a piece of software called DSLR remote which allows you to control your camera by a pc when tethered by a usb or firewire cable. This not only allows you to take and view your pictures, but also to control the settings on the camera such as the aperture, shutter speed and iso. This is great because once setup you don’t have to manually adjust your camera and the images can be seen a few seconds later on a big monitor screen. This is much better than having to peer at the small display on the back of the camera.
The rig was set up so that a falling drop of water would fall through laser beam and so trigger the Smatrig.
I did not want to release the shutter straight away so reading on Luk’s site I saw that he had designed a delay unit. He did not have these for sale but did provide a circuit diagram. I purchased the parts and built it for about £10.
After a short while I had everything set up and working after a fashion. The coloured water drops were falling correctly into a sundae glass full of water in a gardening tray to collect the spills, and I was getting some reasonable results, but having some problems.
The valves that I were using were too industrial. In fact I think that they were meant for water control in railway carriages. They kept burning out the relays which I had to keep replacing.
Looking on the internet I found that the best valves for the job were made by Shako so I sourced them in this country and purchased three of them. I had to buy barbs for them to that I could connect the pipes and fit interchangeable nozzles of different sizes for different sized drops.
The next problem was the lighting.
The water drops are moving at a very high speed having fallen about 500mm and I want to freeze the motion. The motion is frozen by using high speed flash . At full power the flash duration is too long, and the drops suffer from motion blur. By reducing the flash power you reduce the flash duration, and in testing I found I had to reduce the power to 1/64th or even 1/128th to freeze the motion.
This now presents another problem. There is not enough light. So I connected up three speedlights of different makes, but this did not work as they were not perfectly synchronized, and I got motion blur again
Another problem was that my camera after a few shots failed to trigger the speedlights. Back to the drawing board. I tried one flash triggering the others as slaves. The camera liked this but they were not perfectly synchronised. I thought of using a wireless trigger for all the speedlights, but then realised that going down that route would be too expensive.
That led me to design and build my own circuit board using the knowledge that I had gained from building the delay unit (and some help from Luk). This board has 6 outputs and I have used one output to trigger 3 speedlights, so it has the capability of triggering 18 speedlights in perfect synchronisation without the camera protesting. Each bank of speedlights linked together can be tested individually or all at once.
In the mean time I found that other workers were using Yongnuo speedlights so I purchased 6 of them on ebay, and made supports for batches of three on gorilla tripods, so that they could be place anywhere around my setup.
I was now getting enough light to be able to use iso 200 at an aperture of f15, so I was not getting any problems with noise. My camera is a Canon EOS1 Ds MkII which has an image size of 17 Mega pixels. I always shoot in raw and initially open my images up in Adobe camera raw within Photoshop. Here I first sharpen the image and then reduce any introduced noise . Next I adjust the exposure if necessary before opening it in Photoshop for any tidying up, cropping and resizing, adding logo etc.
Here I must mention the camera settings. The camera shutter speed is set at the sync speed of the camera (1/200th second) so that the sensor is not covered by either the first or second shutter curtains. This means that the flash can occur at any time in that 1/200th second window to get an exposure. I have no problems with low ambient light. However if you use a longer exposure time you may have to consider a greater degree of blackout. If you try to use a shutter speed greater than the sync speed then you will get part of the image obscured by the shutter. The aperture is set to give the correct depth of field and the iso is adjusted to get the exposure correct. All done on the pc using DSLR remote. There is no need to touch the camera once set in the correct position and focused
The lens I use is a 105mm f2.8 Sigma macro lens. The camera is mounted on a sturdy tripod with a wind up and down handle, for quickly adjusting the height. As I have mentioned previously the system takes a picture every 10 seconds. I watch the taken picture on my pc and here I can adjust the camera and rewrite the Arduino script if neccessary, and adjust the delay so that the picture is taken at the right time. So you see everything is infinitely adjustable as the system is running. It takes quite some time to get everything set up perfectly but then I can take several hundred shots, each only varying slightly from the next, before making some major adjustments and starting again.
My next problem was that the relay circuitry for controlling the Shako valves was not up to scratch so I designed and built a shield for the Arduino using mosfets , so that now, instead of using relays for the switching , I am using electronic switching.
The push buttons along the top are for purging the valves, and the white button is for testing the camera.
I was far from happy with the sketch that I had written for controlling the Arduino, so I spent some time studying how to do this. Previously all the events happened in sequence with delays in between, and then at the end the whole process started again. The problem with this approach is that events can only happen sequentially, so I could not have more than one event occurring at the same time. No use if you want another nozzle to start forming a drop before the previous one has finished.
I learnt how to use ‘millis’, where a time is set for when an event occurs, and when that time is reached it happens whether it is switching on a pin on the Arduino or switching it off. This controls the switching on and the switching off of the solenoid valves. I also incorporated the switching on and off of the shutter release so as to dispense with the Smatrig and laser beam, and built the circuitry onto my control board.
The Shako valves are connected to the Mariotte syphons by 5mm vinyl tubing and the valves in the water flow system to the Shako valves is made using horticultural irrigation valves and connectors, purchased from B&Q . The supply to the Shako valves can be switched whilst the system is running to change the water colour or composition.
The Shako valves are mounted on a piece of acrylic, with a wide slot cut in it so that the valve position can be adjusted to get the drops to align.
For different applications I made different nozzles by soldering narrow bore copper tubing into the hosetails.
My next problem was with getting the depth of field right. As you know when using a macro lens the depth of field is very shallow, but nevertheless you need the background well away so that the junction between the water and the background is not shown in some shallow angled shots. This necessitated in the building of a long tray that was like an infinity pool, where there is no visible side at the front to obscure the view.
With my experience of using methyl methacrylate in dentistry. (Perspex to you). I was able to build the tray, very easily.
The whole tray is built from clear 5mm acrylic sheeting, glued together with a solvent adhesive. It is about 1000mm long, and 300mm wide. The sides are about 45mm high and the front 30mm, so as to get an infinity pool effect. Drains with taps are placed on the bottom of the tray and in the overflow trough. The whole lot is supported on small blocks of wood and shims to get the levels right.
In the bottom there is a sheet of black glass, and to adjust the depth of the water in the tray, bits of packing can be placed beneath the black glass.
At the back of the tray there is a piece of frosted acrylic that sits into the bottom of the tray. Onto this coloured light is flashed with the speedlights behind it. To get the colours I place coloured gels inside the flash diffuser.
If I am using side lighting I can position the speedlights to the side and flash them through another piece of frosted acrylic. To the other side I have a mirror to bounce light back.
On my rig I also have the opportunity to mount my speedights above the splash tray. I wanted some form of snoot to concentrate the light. I experimented with a few ideas including a bundle of drinking straws, but they were not very successful. Part of my dental equipment was a fibre optic light to direct an intense beam of light to the filling material to cure it. I used this idea to make an attachment to fit on the front of the speedlight . It is made from clear acrylic rod and works very well, aiming a bright flash precisely to where it is required.
Focusing on the correct spot is done by a plumb bob suspended from the centre nozzle. And a piece of a ruler can also be placed here to determine the depth of field when focusing on the plumb bob.
Floating Bells Project
Whilst I was on holiday in Krakov, I was sitting watching the fountain in the main square when got the idea of having an upward spurt of water colliding with a drop descending. When I got home I built the system and got an amazing set of results. Whilst the system was running I was injecting different coloured paints into the flow of the water to give the colourful results. For all of this I was just using one valve to send up the first drop of water and then the second drop collided with this one as it started to descend. The water had guar gum mixed in with it to thicken it slightly. The water was from a pressurised bottle fed from my compressor. The pressure had to be varied dependant on the viscosity of the water and the size of the nozzle being used, so that the spurt was the correct height. The camera was on a tripod with a winding handle to adjust the height if the collisions got too far out of frame.
The lighting was 3 Yongnuos onto the perspex background and 3 Yongnuos from the left side and a mirror reflector on the right.
It was quite a performance because as the system was running I was adjusting the delay to get the image at the right time, the air pressure to get the height right, injecting colour into the pipeline and twiddling the camera up and down on the tripod.
This is what the system now comprises of as a working system.
Nine year old pc with 4GB ram, 2TB hard drive for storage, 110 GB ssd (to speed it up) on which is the operating system and programs. 22” monitor.
Windows xp pro. DSLR remote (to control the tethered camera and receive the images as they are taken). Arduino software for loading up the sketch to the usb tethered Arduino. Windows CS3 and Bridge for viewing the images afterwards.
The Arduino sketches are written on a networked pc which is also used for post production of the images.
3. Arduino micro computer which sends the sends signals to the various digital pins on it according to the installed sketch.
4. Main control panel which interprets these signals to triggering the camera (EOS1 Ds MkII), and provide power to the Shako valves to deliver the water drops.
5. Delay unit which delays the signal to the camera to release the shutter until the water drop collision.
6. Flash control unit, which is needed to trigger multiple speedlights. Also incorporated in this can be another one or two delay units which will give a multiple exposure image if required.
7. Various water containers within an infinity pool type of acrylic tray.
8. Six Yongnou speedlights mounted in two batches of three on small adjustable tripods. Coloured gels , and frosted acrylic diffuser panels
9. Adjustable support system for the four Shako valves and nozzles.
10. Four Mariotte syphons for delivering the coloured water. They are adjustable in height to alter the flow of liquids of different viscosity.
11. Additives for the water. Food colourings, but be careful with the red food dye because after 3 days it will start fermenting in the bottle and leaks out of the breather pipe. This can be cured by adding a few drops of disinfectant to the water. Guar gum for increasing the viscosity. Rinse aid or dishwasher cleaning liquid to reduce the surface tension of the water to get higher splashes.
12 EOS 1DS MkII camera on a sturdy tripod with a Sigma 105mm macro lens.
The development of this system has been a very enjoyable experience and has taught me many new skills. Some of the electronic components that I have developed will find their way into further projects that I intend to follow up. In particular the main control unit, and the flash control unit. And also Luk’s Smatrig and delay unit
Finally I must apologise for all the clutter lying around in my photographs. My excuse is that it is a working environment, and also my garage.
Since I started this project I have become aware of systems that can be bought. You can find them by looking on the internet. They are very expensive, and not as versatile as my system, and to me the greatest pleasure is in designing something and building it. And all the more so when I see my results.
If you want to see this in pdf format for printing, click on the link below