How is the water effect achieved in “Dali Atomicus”?

Question

I've been thinking of doing a reinterpretation of the Dali Atomicus by Philippe Halsman, and this curved water effect got me puzzled, so I wonder how this is done? If I splash some water from a bucket I manage it to go only downwards in a single curve, but this has (at least) two curves, if looking from left to right, upward to a straight and slight downward.

Answer 1

As Stan notes, apparent water path can be substantially affected by the motion of the "bucket" or other water source. If you can move the water source rapidly compared to the velocity of the ejected water you can achieve complex shapes.

The composite photo below shows a number of attempts to achieve a version of the much tried summer-beach-hair-water-cliche photo. It's been done a million times, but can still be fun :-). (Trying to synchronise 2 people makes it far far harder).

As can be seen in the multiple images, the water source (hair) is not where most of the water is when the photos is taken, the paths can be given complex curves, and the water is relatively stationary in the air when the photo is taken.

---

How far in how long":

Distance travelled under gravity when starting from rest is given by
distance = 0.5 x a x t x t = 0.5at^2. Using SI system: a = gravitational acceleration ~= 5 m/s/s, t = time in seconds and distance is in metres.
OR distance = 5 x t^2 metres. So

        time, seconds  
          |           fall  
          |             |
         0.1 s ...   50 mm   
         0.2 s ...  200 mm  
         0.5 s ... 1250 mm  
         1 s ..... 5 metres !!!   

This means that if you can take the photo very soon after the water has been deposited, very little motion will occur due to falling.

How fast?:

Speed of fall affects sharpness.
Speed is more forgiving than distance. Speed = gravitational acceleration x time.
Essentially, speed of fall
~= 5 x time in m/s.
~= 18 x time kph ~= 11 x time in mph

Answer 2

It's just a matter of moving the bucket correctly. It looks to me as though the trick is to make the initial "toss" at about waist height (from camera left), then move the bucket quickly down and and back while the momentum carries the water forward. You may not be able to duplicate it exactly, but I don't think the shape of the bottom of the arc was planned exactly anyway -- it would mostly have been about getting the water-wrangler's hands and arms out of frame in time. Unlike the chair holder (whose arm was removed in retouching), the water guy's arms would have blocked a part of the scene that would have been difficult to retouch properly, especially if it had occluded the turned front leg of the chair.