I do not know much about creating movies or the special effects associated with them, but I can imagine a scene from the 70s when the first Star Wars movie was made. In one scene, they have to show an escape pod that moves away from a rebel blockade runner with the out-of-control droids R2-D2 and C-3PO. Things are tight, so they have to do something fast.
What to do? According to a modeler in this video (and as described in this post), the Star Wars crew built the escape pod out of a Kentucky Fried Chicken bucket and dropped it from a height of 1
But is it really what happened? Ah ha! We do not have to take that for granted, we can actually test this story. If I can measure the movement of the escape capsule in the Star Wars scene, I should at least be able to tell if the model has accelerated. If it moves at a constant speed, it must have produced the effect in a different way.
Here is the plan. Find the position of the model as a function of time. An accelerating object would have a position-time graph in the form of a parabola. So I only need the position data. If I could see the "falling" model from the side, I could easily use video analytics to get position-time data by looking at every image in the scene.
Unfortunately I do not have this view. The lack of side view makes this analysis challenging and interesting. If it were easy, it would not be so much fun. Why do not more people say "unfortunately"? It seems like a great word.
OK, I still can determine the distance from the model to the camera by looking at the angular size of the image. Angle size is a measure of how tall the object looks rather than its actual size. If you need an example, extend your thumb and place it between your eye and another person's head. Your thumb looks bigger than that person's head, but obviously you do not have a huge thumb (I suppose).
When an object is sufficiently far away, the object's length is approximately the arc length of A circle that passes through the object when centered on the viewer. This diagram should help here.
This relationship between distance, angle size, and length only works if the angle size is measured in radians (ie, an angle size of 2π would match the circumference of a) circle).
We do the following. Take the scene with the moving escape pod. Suppose it is a chicken bucket with a diameter of about 0.2 meters. Mark the position of two points on the escape cage (I still can not overcome the KFC thing) and find the distance between those points. Oh, I have to assume an angular field of view for the camera – I'll just pick 1 radian to make it easy. Once I found the "distance" between these points, it would give the angle size. Then I can determine the distance to the camera based on the angle size and the actual size.
After that, I get the following distance from the camera to the KFC model.
Um … that looks like a parable to me. If I adapt an equation to this data, I get an acceleration of 1.65 m / s 2 – and I do not expect that. The acceleration of a free-falling object (on Earth) should be 9.8 m / s 2 . Does that mean that the video is fake? Of course it's a fake, it's a movie. It is not a real life. You probably just slowed down the playback frame rate to make it look cooler. If you change the frame rate from 24 frames per second to 58.5 fps, the correct acceleration should be achieved. Of course, it could look weird for a rescue pod. But I'm pretty sure this is just a falling KFC bucket that looks like a spaceship – just like the guy in the video said.
Other great WIRED stories