Friday, January 6, 2017

Water Droplet Photography

Like astrophotography, high speed imaging is another intersection of technology and photography. I wanted to try out some basic high speed photography, and last night I took photos of water drops falling into a bowl of water.


The general concept behind high speed photography is that instead of triggering the shutter on the action, which is very hard to do when the moment only lasts thousandths of a second long or shorter, a light is flashed at the moment of action in dark setting. This effectively "freezes" the moment you want the camera to see since . A flash bulb normally used in photography can provide a pulse of light lasting a thousandth of a second, and is suitable of high speed photos of water droplets. Air-gap flashes can provide incredibly bright flashes that last under a microsecond, freezing the fastest of motion without any motion blur.

Using what I had on hand, I created a simple circuit to flash a bank of LEDs instead of trigger a flash (since I didn't have one). 


LEDs are easy to use because I can just switch them on and off with a MOSFET. I have no idea what the rise and fall times are on the LEDs I'm using, and the LEDs are mounted together as modules I bought over three years ago so I don't know what the on-board circuitry is doing, but it is most definitely not going to improve response times. However, I'm only imaging falling objects, not a bullet, so a rather long flash should be fine.
LED module, originally used in a robot.
The hardware is straightforward: a photoresistor was set up in a voltage divider, and a N-channel MOSFET was used to switch the LED bank on and off. I used three modules as the LED bank for the flash. A LED with a heatshrink shroud is turned on at all times and aims into the photoresistor, creating a gate.

The Arduino code is very simple; all it does is wait until something passes through the photogate, delays for a bit, and then flashes the LEDs for half a millisecond. Not a very short exposure, but I without it the scene would be too dim.

I set up the whole thing in a dark room at night. As you can see in the earlier photo, the setup is rather hacked together but it works. I used a mixture of water and milk as the liquid in the base and for dropping.

The exposure settings were 2-4 seconds, f/6.3, 400 ISO.

A quick calculation of the time to impact of an object falling from rest 20 cm up gave around 0.2 seconds. I initially did this in my head and thought the square root of 0.04 was 0.02 without realizing how wrong that was. I spent ten minutes trying to figure out why I couldn't see anything in my photos. I dropped multiple things in trying to just see where it was. After fixing my error; I got something:
Chopstick.
With timing sorted out, it was time to get the photos I originally sought. I didn't have a pipette, dropper, or syringe on hand so I first tried making drops with a straw. Just like when I was kid, I captured the liquid in the straw with my thumb sealing off the top end, and then let the liquid loose by releasing my thumb.

It was quite hard to create a single drop with this method and many times I just released a mini stream. My dad had an idea to create drops by dipping a chopstick into the liquid and then just raising the end of the chopstick just over the photogate. The liquid along the sides of the stick run down and pool into a drop that falls off. He helped me make the drops while I set off the shutter.

Success!
Finally!

By varying the delay and the drop's initial velocity, we could capture different parts of the splash.
Impact!
Crater.
Liftoff.
In the future, I need to better isolate the light used in the photogate. I don't have them on hand right now, but an IR LED and photodiode would work well here. To further reduce ambient light, I will probably add on a fourth LED bank to make the photo brighter and shorten the exposure so the background light has less time to integrate. I would also get to use a lower ISO and smaller aperture with a brighter flash, bringing more of the image into focus and decreasing noise. I can also use a shorter flash with the brighter light to reduce blur and capture faster objects. Currently, without the flash of the LEDs, the background is relatively bright:
Okay, not that bright, but still visible.
To really shorten the exposure would involve electronically controlling the shutter so that it is open just before the flash and closes just after.

Also, I want to try pulsing the light like a strobe so I capture multiple instants in a single exposure. Should be interesting.

Update (1/7/17):
After a bit of googling I found a blog that talks about the lighting modules I used from Digitron. I won't go over all of the post, but it did cover the response time of the LED module. The little five pin SMT device on the board is an LED driver, the ZXLD1350. Instead of toggling power to the modules, the author toggled the adjust pin on the driver IC. They got times of 110µs to turn on and 400µs to turn off. My program for generating the flash turned on the module for half a millisecond, or 500µs. Clearly, the response time of the modules is going to lengthen the light pulse and cause blurriness. However, in my circuit I am not controlling the driver but applying and removing power to the whole module. I'm not sure how the response time is affected by that.

At worst, the pulse can be twice as long as intended, causing motion blur. I'm going to need to find a better setup to get higher brightness and faster response times. The Vela One is a commercial LED flash for high-speed photography and boasts a pulse time as short as 1/2,000,000 sec (0.5µs). I'll likely never reach that level (or ever require a pulse that fast), but let's see how short I can go. Time to go on the search for new circuit setups and LEDs.

Also, half my modules don't have the driver circuit:
Middle module doesn't have the driver and supporting components while the right one does.
Datasheets: module and driver

No comments:

Post a Comment