Is This Knife Ping Pong Video Real or Fake?

I’m not sure what the internet says about this video (I don’t want to look yet). However, this is another great example of a video that could be real or fake. Let’s take a look a couple of things in the video that will help us determine if it is a fake.

Looking at Camera Shake

This is an older trick, but it might still be in this video. The idea is that a video editor would like to add some special effects to the video (maybe a computer generated image (CGI) knife – maybe). Of course if you add a CGI knife to a moving background you need to also move the CGI knife. Wouldn’t it be easier to have a stable non-moving background? Yes. You could easily record a video using a tripod and then add your effect.

However, using a tripod could be a dead give away that you are adding CGI effects. So, the trick is to record using a tripod and then add a fake camera shake after you have added your CGI effects. Boom.

In some cases, it is trivial to show a fake shake. Also, it is fairly easy to check for this type of effect. This is almost always the first fake video test that I do. Here’s how to look for this in a video.

First you need to download the video and also get a video analysis program like the free Tracker Video Analysis. Now all you need to do is to track some object in the background of the video. You don’t even need to scale it or anything.

For this particular video, I can pick the corner of that window frame to the right of the pink girl. If you choose carefully, you should be able to pick a background feature that will allow you to use Tracker’s autotracker feature. Trust me, this is a helpful thing.

Here is a plot of the trajectory (x vs. y) for the first few seconds of the ping pong knife video.

Background motion from Ping Pong Knife video

Is that fake? What should it look like? Hold on for a second. Let’s take a look at just the horizontal position of the background as a function of time.

x-position vs. time for background motion of Ping Pong Knife video

That looks a little suspicious – a little too smooth and regular. What about the vertical motion?

y-position vs. time for background motion of Ping Pong Knife video

It essentially looks the same. What about a video that I know is real? Here is a video I just recorded. I held my phone with two hands and propped my elbows against my chest to make it as steady as possible.

Here are the same plots of x-y, x-t and y-t for this real video.

Background motions for my real video

In this real video, the trajectory plot looks more like a random walk than the ping pong knife video trajectory. Also, the x-t and y-t plots in the ping pong video look very similar. However, in my real video these two plots do not look similar.

I think there are many possible ways to analyze this background motion (I have used frequency analysis before). But let’s look at the horizontal velocity. This is the velocity vs. time for the ping pong and real videos.

Comparing real background velocity with Ping Pong Knife video

Notice that the real video has a velocity that jumps around but the ping pong video is very smooth. Usually, a velocity graph is going to be jumpy since small variations in position can lead to much larger variations in velocity. Maybe the ping pong video used some post-video smoothing process. Youtube has this editing feature built in for uploaded videos. Here is a plot of the horizontal velocity after applying youtube’s “steady” filter.

Background velocity for real video after youtube stabilization filter

It still looks jumpy. The ping pong video still looks suspicious.

Accelerating Objects

There is another thing to look at in a video – the acceleration of falling objects. The last knife throw chops that pineapple in half. Let’s take a look at the falling piece after the cut. Now, there is one problem here – scale. I need to know the scale of the video. I am just going to take a guess that the pink girl is about 5 foot 2 inches. Actually, this would make the ping pong paddle about 15 cm across which seems to be a standard.

Here is a plot of the top half of the falling pineapple.

Acceleration of falling pineapple piece

There isn’t anything conclusive about this motion. Yes, the vertical acceleration isn’t 9.8 m/s² but that could be because of a scale error or a perspective error (the pineapple moves away as it falls). It would be nice if there was a sideway view of the knife throw of the pineapple.

Conclusion

As with any video, there is always a judgment call. For this particular video, I am going to go with “fake.” There are two reasons for this decision. First, the background shake doesn’t seem correct. Second, the stunt seems excessively dangerous, along with the difficulty of the knife hit. I’m not sure it’s so easy to hit a knife with a ping pong paddle, much less get the knife to chop a pineapple in half. If there was one small mistake, the pink girl would have a knife in her head (and this is bad).

Of course, I have been wrong before about real and fake videos. Very wrong. Oh, I love the guy coming down the stairs at the end. He looks real.

Bonus Image

I was going back through the video and looking at the knife trajectory to see if I could get an analysis.  Take a look at this frame (I added the two red arrows).

Frame from Ping Pong Knife video

The arrow on the right points to the thrown knife.  You can go frame by frame and check that it is indeed the thrown knife.  The arrow on the left points to the arrow stuck in the paddle.  Yes.  The knife is both in the air and stuck in the paddle at the same time.  It’s sort of like the knife is in two quantum states at the same time.  It’s both stuck and moving until you observe it.  Oh wait, I just observed it – it should either be stuck or in the air.  I guess the video is fake.

Source: http://feeds.wired.com/c/35185/f/661500/s/3249ea12/sc/4/l/0L0Swired0N0Cwiredscience0C20A130C10A0Cis0Ethis0Eknife0Eping0Epong0Evideo0Ereal0Eor0Efake0C/story01.htm
Related Topics: denver broncos   GTA 5 Cheats   What Does the Fox Say   Vma Miley Cyrus   amanda bynes