• Home
• Forums
• About Us
• Blog
• Contact
• Links

Getting the most from your Tennis Video

The other day, Will and I were discussing how much buzz there seemed to be surrounding tennis video, and how many people wanted to share video of their strokes online. Not only that, but many people also seemed to be interested in using video analysis to help them improve their game. So I had an idea to write an article that would specifically help people answer the question: "How do I make a video of myself or someone else playing tennis, starting with no knowledge of cameras or software?" And I will try to do it in a plain-English way that makes sense whether you're a pro photographer or not.

In this article, we will take a look at how to get the most out of your tennis video. I will attempt to "cover all the bases," so to speak, and give you an understanding of how video works, and how to operate a camera. Then I will discuss how to shoot, view and edit your material, and finally how to share it, all using FREE software. And of course, I will relate it all to producing a better tennis video. This will be a several part series since there's a lot of information I want to convey!

We will also include video clips on the website for all of you visual learners out there.

Part 1: Understanding Video: Frame Rates and Interlaced vs. Progressive video formats.

Let's start by getting an understanding of what video is and how it works. In very simple terms, video is a series of pictures, and each picture changes to the next picture quickly enough that it tricks your eye into thinking that you are seeing real motion. You may have seen a similar effect if you ever owned one of those cartoon animation books, where you would use your thumb to flip through the pages of the book so fast that the drawings on the pages looked like they were an animated cartoon. When you watch television, your TV is changing the picture 30 times every second. When you go to the movies the picture is changed slightly more slowly, about 24 times per second. The newest generation of HDTV sets on the market can change the picture 60 times per second, so there's certainly some variation across all these different formats. But again, changing the picture many times every second tricks your eye and brain into believing that there is motion going on, even though you're being shown a rapid succession of still pictures. Each different still picture in a video is called a "frame."

The number of times the frame changes each second in a video is called the "frame rate," and it is abbreviated FPS for "frames per second." Most consumer video cameras shoot at 30 frames per second (30fps), and as I mentioned above, most television is delivered to you at 30fps as well.

Slow Motion

Now, even though most cameras record 30 pictures every second, you can change the speed at which you play back the recorded footage. If you play the footage back at say, a speed of 10 frames per second, you get slow-motion, because it takes you longer to watch than the original recording took to occur.

Let's use a tennis serve to illustrate slow-motion in video. If you pointed a video camera at someone serving, it might take the player 1 second to go through his entire service motion and hit the ball. During that 1 second, the camera would capture 30 individual frames. If you played those 30 frames back at the normal 30 frames-per-second rate, you would get normal speed, "real-time" playback. But if you played the video back at a rate of only 10 frames per second, it would take 3 seconds for you to watch a serve that actually occurred over a 1 second time period. Again, this is what we call slow-motion.

Some specialized cameras can capture video at upwards of 1000 frames per second, and when you take that recorded video and play it back at a normal 30 frames per second, the result is what we call super-slow-motion video. Again, let's use a person serving as an example. If it took 1 second to serve but you captured 1000 frames during that second, and you then played the footage back at the normal video rate of 30 frames per second, it would take 33.33 seconds to watch. To get an idea of what that kind of super-slow-motion looks like, check out this clip of former Davis Cup player Oliver Akli hitting his forehand.

Key points:

• Video is made up of a succession of pictures, more properly called "frames."
• The number of frames recorded or played back is called "frames per second" or "fps."
• Most video cameras shoot 30fps, and almost all TVs play back at 30fps.
• Playing back video at a slower frame rate than it was captured at results in slow-mo.

Resources:

http://en.wikipedia.org/wiki/Frame_rate

Wikipedia article on frame rates. Useful information, although it focuses too much on how frame rates affect video gaming.

Understanding Interlacing

In any discussion of video, and especially sports videography, there is one more technical detail that must be covered: "interlacing," It is important to understand how interlacing affects cameras and your tennis video.

What the heck is interlacing? It turns out that it's not exactly true that an average TV shows you 30 frames a second and that an average consumer camcorder records at 30 frames a second. Most TVs and camcorders handle a video signal a little differently.

Decades ago in the 1930s when the first video cameras and TVs came out, the TV sets weren't very "technically capable" (keep in mind that video is different from film, which has been around for over a century). These first TVs didn't have the electronics horsepower to actually display 30 full frames each second, so the engineers settled for what you might call a "hack job." Instead of actually displaying 30 full frames each second, they would display 60 half-frames each second. Each half-frame is called a "field." This doesn't mean that the top half of the screen is one picture, and the bottom half is another. The two images alternate rows horizontally down the picture. Although this sounds strange, and although it also sounds like the same amount of work as displaying 30 full frames each second, is actually much easier to do.

Here's a practical example of how interlacing works: A normal TV signal is made up of 480 horizontal lines of resolution (HDTV has up to 1080 lines). So when you look at a TV screen what you're really seeing is 480 lines of different colors and brightness which, from the comfort of your couch, end up looking to your brain like a picture.

In an ideal, non-interlaced world, all 480 of these lines would be drawn simultaneously 30 times each second. More on that later.

Let's say a television starts to receive an interlaced signal. The first field comes up, which again is a frame but with every other line cut out. So the field only actually has information for the odd-numbered lines, lines 1, 3, 5, 7, and so on. The TV takes that information and "paints" it onto the screen on lines 1, 3, 5, 7 etc.

1/60th of a second later, the second field comes along in the signal, and the second field only has information for lines 2, 4, 6, 8, and so on. The TV then paints the second set of lines, the even numbers, onto the screen. Now that both the odd numbered lines and the even numbered lines are painted onto the screen and 1/30th of a second has passed, we can call this combination of Field 1 and Field 2 "Frame 1" even though frame 1 is a combination of two different pictures sharing the screen at the same time.

Another 1/60th of a second later the third field comes along, and that field contains new information for the odd numbered lines 1, 3, 5 etc. The TV paints the new odd numbered lines, but leaves the even numbered lines alone. Another 1/60th of a second later, field 4 arrives, containing new information for the even lines. The TV then paints those new even lines, but leaves the odd ones untouched until frame 5 arrives, etc. The end result is that the entire picture does in fact change 30 times each second, but at no point in time is a complete, actual "picture" shown on the screen. Each picture is always sharing half of the screen (every other line) with the next picture in the video sequence.

A camera recording an interlaced signal does the same thing. It opens its electronic "eye" and records a picture, but it only records the odd-numbered lines. Then 1/60 of a second later, it opens its eye again and captures information to write to the even-numbered lines. This is where the problems for tennis video come in. Let's say you swing your racket extremely fast. The camera opens its electronic eye and snaps field 1, which it writes to the odd-numbered lines. Then 1/60 of a second later, it takes a snap and writes that information to the even lines. But in that 1/60 of a second, your racket has actually moved a significant distance because you're swinging fast. When you play back the video and try and freeze it, you end up with a picture where half of your racket is a foot behind the other half, and it is very visually confusing.

Why in the heck did anyone ever use this system of interlacing? Again as I said before, this is 1930s technology, and it is actually a very clever analog way to double the amount of visual information on screen and double the rate at which things change onscreen (smoothing out the appearance of the video to the eye) without actually increasing the bandwidth required to send and receive the picture.

Because this technology developed for decades and became the standard, we are still stuck with it today. The technology to overcome having to interlace, called a "frame buffer" did not become available until the 1980s, so television and camera systems developed for 50 years using interlaced technology, and it's hard to "shake that off," so to speak.

Progressive Video

Many consumer video cameras today can only shoot and record an interlaced signal. Thankfully, some higher-end consumer cameras have the digital processing horsepower to shoot what is called "progressive" format video, where each frame is captured and recorded in its entirety. This is the ideal format for capturing video, especially sports video. You may have heard this term when shopping for a DVD player at a store like Best Buy. A "progressive scan" DVD player sends a progressive signal to your TV, allowing each frame of a movie to be seen fully rather than interlacing the signal. If your TV can display a progressive signal (most can as of 2008) the result is a movie experience that looks much more "film-like" than an interlaced signal.

If the video camera you already own only records interlaced signals, you are sort of out of luck, although you can use computer software to change interlaced video into progressive. This type of software generally works by taking the first field, which contains the odd-line information, and simply "copying" that information and placing it into the empty even lines, giving you a complete first frame. The process is then repeated for each frame after that. This technique is called "line doubling." There are other techniques of course, some of which involve the computer attempting to interpret what the missing field lines should look like based on what the visible lines look like.

Astute readers or people knowledgeable about video might be wondering, then: If you can use line-doubling like that, then can't any camera that shoots 60 interlaced fields per second actually be tricked using software into creating 60 full progressive frames each second, and thus "higher-speed video?" The answer is "Yes!" But the problem is that each one of those 60 progressive frames really only contains half of the picture information, merely doubled over. So you've chopped the true resolution of the video in half by doing this. In fact, some cameras can do this hack-job internally. Our Canon GL-1 camera that we use for some spot shoots has what it calls "frame mode" in which it turns its normal 60 interlaced signal into a 30 progressive, but the results are pretty abysmal in terms of quality (it is still suitable for web use, though). The higher-end Canon XL2 cameras that we use for the majority of our material can record in a "true 30-progressive" format. These cameras can also record 60- interlaced signals as well as 24-progressive signals, giving us greater flexibility.

One final note on frame rates, interlaced and progressive video: Usually, a video signal is referred to or abbreviated by three things.

1. The number of horizontal lines of resolution in the signal (480 for standard definition, 720 or 1080 for high-definition formats)
2. The frame rate of the video (usually 60 if interlaced or 30 if progressive, although 60 progressive is avant-garde as of 2008)
3. Whether it is an interlaced or progressive signal.

So a typical camcorder or television signal we would refer to as "60i" because it is 60 interlaced fields per second. 30p means 30 progressive full frames per second. 24p is 24 progressive frames a second, which is the same frame rate as cinema film (so 24p has that "film" look to it).

Key Points:

• Interlaced video uses half of two separate frames to make one image
• Progressive video signals are a true "series of pictures" or frames
• In reality, many cameras record an interlaced signal, not a progressive signal.
• Progressive is better, because it avoids issues that occur when objects move quickly on screen, such as a tennis racket or ball.

Here's a two-question pop quiz. If you answer these correctly you're well on your way to understanding how to make a better tennis video.

Q: If I told you that I watched the US Open in 720/60p format on the CBS HD channel, would you know what I meant?

A: 720/60p describes to video signal coming into my TV set. 720 means that there are 720 horizontal lines of resolution in the picture, which is an HDTV format. The "p" means that the signal is progressive, so I am seeing a signal that contains full frames, not two interlaced fields. The "60" means that the entire screen is being re-drawn 60 times per second. Sounds like a great way to watch some tennis!

Q: Would you rather watch the US Open in 1080/60i or 720/30p?

A: This is a tricky question! Both 1080i/60i and 720/30p are high definition video formats. The 1080/60i signal is actually higher resolution than the 720/30p signal (1080 lines instead of 720!). But again, let's remember that at any time on the screen with the 1080i signal, you're seeing two separate images interlaced together. The two images share the screen, so each gets only 540 lines to itself; less lines than a full frame gets in a 720p signal. The 1080/60i signal will look smoother on screen because the image is changing twice as fast as the 720/30p signal.

Bottom line? The 1080i signal will be a little smoother and more realistic looking, and actually packs more visual information onto the screen (higher resolution). But the 720p signal is progressive, so you get a more "pure" picture experience. Unless you are a video purist, you may not notice the difference while watching TV. But most sports video aficionados will choose 720p every time.

Q: Would anyone out there like to buy Fuzzy Yellow Balls two video cameras that can shoot high definition, true 720/60p? Just kidding!

But seriously, you know you want to. They're only $20,000 each when you include the lenses.

Resources to read further:

http://www.100fps.com/

An excellent site that explains why deinterlacing still haunts us today (its 1930's technology!) and has tons of example images. Also shows you how to use software tricks as I mentioned before to get higher frame rates from your camera. One other note: 100fps.com is based in Europe. Over there, the standard (called PAL) is that camcorders and TVs work at 25fps (50i), so don't let that confuse you with the American standard, called NTSC, which works at 60i as I have previously explained.

Read Part 2 »