In this post i will try to shed light on how to stabilize your shot. In other words how to remove the parasitic movement from your still or dynamic shot.
This post could be really long and I will make sure to update it often.
I will first give a general (brief) overview of what is stabilization .Then will divide the problem in two. First I will talk about the orientation problem . Second I will introduce the translation problem.
What is stabilization?
Stabilization is defining for a system a set of parameter to work in and providing the mean to stay within this range despite unpredictable perturbation. It fall under general "theory of control" in system engineering. Common application are assisted steering , mechanical arms , AC ,etc...
For our cameras, that is defining an aim (what we point at ) and ensuring that we won't deviate too much or have unpleasant shakiness on the screen.
Imagine a ball on a track having to stay in one position . We will call it equilibrium .
if it was on a hill , the ball would be unstable. WInd or any other perturbation could make it fall of the cliff.
Let 's take the same ball on a different track shaped like a valley.
It's important to notice that for the purpose of photography , we will prefer smoothness other speed, and responsiveness other accuracy, and avoid oscillation at all cost.
The orientation problem
The notion of orientation is pretty intuitive to us. For our camera it is what we point at, and for us and our vehicle it is what where go . Generally it is the direction of the shortest path to an object or location relatively to our own position. As you may have realised it is not the easiest way to manipulate our orientation or communicate it to others.
As you already know the physical space in which any object around u evolve is in 3 dimensions. It is pretty intuitive to think about going up an down , left and right and forward and backward.
It turns out there are similar way we can describe our orientation. In the industry of photography we will refer to rotation on the 3 main axis as tilt,roll and pan. they are similar to the terminology of yaw,pitch and roll, or the x,y,z used by the scientific community.
Ba, Sileye & Odobez, Jean-Marc. (2008). Recognizing Visual Focus of Attention From Head Pose in Natural Meetings. IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society. 39. 16-33. 10.1109/TSMCB.2008.927274.
If you are familiar with the classification of camera shot you won't have any problem remembering those terms. It's important to note that multiple sets of pan, tilt,roll can result to the same orientation.
How does that affect my shot ?
The impact of small variation of orientation in your shot is relative to the distance between your objective and your object. The further an object is from the objective the bigger the displacement will be in your shot .What are the solutions?
How does it work?
To understand the steadicam it is important to understand why object roll in the first place?
"Objects roll to lower their center of gravity".
In short the same reason why they fall. I know that's seems extremely obvious , but it is important to remember. If you don't believe me watch this
In short a Steady Cam add a stick down you camera and artificially transfer the center of gravity( beginning of the arrow ) way low down the stick by adding weighted plates .
Why does it works ?
Steadicams come with a floating handle which serve as center of rotation for the apparatus whole camera and stick. Just like a long pendulum the center of weight of a steadicam is in a stable equilibrium.
Pros and cons
- + The steadicam is an entirely passive device (no energy consumption).
- - The main drawbacks of a steady cam is its weight. It result can be tiring to operate and its make it hard to combine with other systems.
Modern gimbal are generally battery operated device relying on 3 servo motors to maintain a load orientation. They are inherited from the fields of robotics and have similarities with industrial mechanical arm.
How does it works
Gimbals need you to balance your camera on the roll and tilt axis. Gimbals once balanced have generally at least 3 modes of operation. They each correspond to the degree of freedom given to the load (camera).
- Locked: in locked mode all 3 angle (pan,tilt, adn roll) are locked in position.It allows the operator to move freely the camera without impacting its aim. This mode is ideal for follow and trucking shot , or while still shooting in a vehicle.
- Pan free: in Pan free mode the gimbal will let the camera follow the panning angle given manually by the operator.
- Pan and tilt free: Same thing but with tilt additionally subject to the motion of the operator.
Why does it work?
Gimbals are equipped with a combination of accelerometer and gyroscope sensor called in robotic an Inertial Measure Unit (IMU). Those IMU are used to track orientation at relatively high frequency compare to a human operator. That allow them to be both very precise and very responsive.
Gimbals need you to balance their loads(camera +accessories). To help you reach that goal you can use counterweight that you can find here.
Since their load is balance they only need to stabilize it locally. To once again use the analogy of the ball on a track, they don't need to invert the hill into a valley but just add bump to it. For this reason gimbals might not react well to extreme angles and you should always experiment first before you try a manoeuver during your shot.
Gimbals need you to balance their loads (camera and accessories). They have maximum weight and volume ratings. It is important to pay attention to those when buying one.
Pros and Cons
- + Fast and accurate
- - Expensive
The translation Problem
A translation is the most natural motion our mind can understand. However the special case of rectilinear translation is what come to most people mind when hearing this term. There are multiple way we can define this motion , however we will use one that build on the knowledge we accumulated.
"A translation is any movement preserving the orientation of an object"
That was easy. Now let's be more specific to the field of photography. As we walk in turn out that our body is not moving in any form of translation. When we think about it none of our body part are. Most of our bones are connected to a dominant joint and are rotating around different axis. Those combined rotation result in a motion of our upper body when we walk called bobbing. It's literally the fact that our head and shoulder are moving up and down. When holding a camera this parasitic motion can often be seen on the footage and is rather uncomfortable for the audience.
The Bob dampener
There are multiple device that aim to compensate the bobbing motion.One common architecture is to use a collapsible parallelogram or also known in Mechanics as a four bar linkage.
Fig a. A collapsible parallelogram
How it works?
Parallelogram have many interesting properties , among them:
- their opposite side stay always parallel
- there dimension do not define them uniquely
As you can see in Fig a.:
- allow us to control the direction of AH by controlling DG and hence use them as handle and camera support.
- Given the right condition its possible to move DG ( the camera support) in height almost independently from AH (the handle) .
It is important to realize that D and A are still connected. The attentive reader will notice that in fact D is rotating around A. This imply that in order for A to be stationary in elevation the horizontal distance with D will have to fluctuate. It could be either the camera moving horizontally or the swift hand of the operator.
Why it works ?
Let's take a closer look at our mechanism.
- In our first animation we can see that for A (our camera) to move up our parallelogram need to maintain its shape. Specifically HD need to be constant.
- In our second animation we can notice that for A (our camera) to stay still (in elevation), the parallelogram nee d to expand. Specifically HD need to expand.
Both of those alternative cost energy. Nature being energy conservative will at each moment choose the easiest. We can notice a competition between the gravitational potential energy and the energy stored in the spring.
In other words , if HD (the diagonal) can expand with less effort than it take to lift the mass at A (the camera), then A will not increase in heigth.
We succesfullly found a condition on the spring place in HD.
The spring need to be relax enough to allow the parallelogram to shift to achieve stabilization .
Let's look at the downward motion.
Similarly moving down the point D while maintaining A( the camera) imply shortening the distance HD.
From an energy perpsective, lowering the point A imply losing gravity potential energy , on the other hand shortening HD imply releasing energy from the spring.
There is then a transfer of energy between the spring to sustain the potentiel energy (height) of A (the camera).
We then make the prediction:
- A spring too stiff will keep the shape of the parallelogram and transmit the motion upward.
- A spring too lax won't allow keep enoug tension and transmit the motion downward.
It's seems that tuning is an important factor.
Pros and Cons
- - It seems that the single parallelogram dampener wil generate horizontal motion
- + the dual handle seems to transfer that problem to the operator