Collect the Mets

The light and camera it takes to freeze the action

Last time, we looked at how to snap a picture with the ball over home plate, which turned out to be much easier than the numbers would suggest. Now though, we have a tougher challenge: making that ball stop moving. Let’s take a look at the picture we started with at original size.

The ball is in the right place, but it’s in a bit more of the right place than it should be. What went wrong? Well, from our calculations last time, we know that the ball is moving at around 1400 inches per second. The shutter speed in this photo was 1/1600s, so that tells us that the ball moved approximately .88 inches in the time that the shutter was open. A baseball is about 3 inches in diameter, so it had enough time to move almost one-third of that distance. That just won’t do. What would we get for different shutter speeds?

Distance traveled during common shutter speeds

V	V(in/s)	1/500	1/1000	1/1600	1/2000	1/2500	1/3200	1/6400	1/8000
70	1232	2.46in	1.23in	0.77in	0.67in	0.49in	0.39in	0.19in	0.15in
80	1408	2.82in	1.41in	0.88in	0.70in	0.56in	0.44in	0.22in	0.18in
90	1584	3.17in	1.58in	0.99in	0.79in	0.63in	0.50in	0.25in	0.20in
100	1760	3.52in	1.76in	1.10in	0.88in	0.70in	0.55in	0.28in	0.22in

And this tells us, well, nothing on its own. The shutter speed you’ll need to freeze motion of a given speed is dependent on the sensor resolution, field of view, and distance to the subject. We can cheat a little bit if we have a reference to work from in the plane of motion and in this case we do – the batter’s box. Or, more accurately, both batter’s boxes.

Assumptions:

• The batter’s box is 72 inches from front to back.
• Image is from a Canon 5D Mk. II (21 megapixels, 5616 x 3744).
• Lens is a Canon EF 70-200mm f/2.8L IS USM at 200mm with a Kenko Teleplus Pro 300 AF 1.4x teleconverter.
• Distance from subject is approximately 100 feet.
• Offset between focal plane and plane of motion is approximately 7 degrees.

In the original of this picture, the outside edge of the near side batter’s box is approximately 2600 pixels long. The outside edge of the far side batter’s box is approximately 2400 pixels long. This would make a 72-inch path down the middle of the plate 2500 pixels in this case. Note that the focal plane here isn’t perfectly aligned with the plane of the batter’s boxes, but that is not a factor in our calculations. If we were to account for this offset in calculating the distance in the focal plane, we would just have to take it out again when looking at an object moving in the reference plane. In any case, we now know that the path of the ball in this picture measures in at approximately 34.7 pixels per inch (2500 pixels / 72 inches). The .88 inches of movement in a 1/1600s exposure should result in 30 pixels of blurring. We can check our work by superimposing a 30 pixel reference on the original image like so:

And that looks about right, give or take a couple of pixels (or +/- 3.8 pixels for the full 70-90mph range of expected ball speeds we determined last time). So far, so good. This tells us why the ball was blurred so much, so how do we stop it? We can assume that motion is frozen if there is less than one pixel of movement in the time that the shutter is open. Let’s look at the exposure distance chart again with the distances converted to pixels.

Pixels traveled during common shutter speeds

V	V(in/s)	1/500	1/1000	1/1600	1/2000	1/2500	1/3200	1/6400	1/8000
70	1232	85.6px	42.8px	26.7px	21.4px	17.1px	13.4px	6.7px	5.3px
80	1408	97.8px	48.9px	30.6px	24.4px	19.6px	15.3px	7.6px	6.1px
90	1584	110.0px	55.0px	34.4px	27.5px	22.0px	17.2px	8.6px	6.9px
100	1760	122.2px	61.1px	38.2px	30.6px	24.4px	19.1px	9.5px	7.6px

Even at the typical SLR’s maximum shutter speed of 1/8000s, there’s still going to be about 6 pixels of movement. To get that number down to one, we’ll have to go to extremes.

Pixels traveled during extremely fast shutter speeds

V	V(in/s)	1/10000	1/25000	1/50000	1/100000
70	1232	4.3px	1.7px	0.9px	0.4px
80	1408	4.9px	2.0px	1.0px	0.5px
90	1584	5.5px	2.2px	1.1px	0.6px
100	1760	6.1px	2.4px	1.2px	0.6px

1/50,000s. One fifty-thousandth of a second. Believe it or not, this is not impossible to achieve, though it can be a bit tricky. You won’t get there with a mechanical shutter, but some CCDs can support an electronic shutter of up to 1/100,000s. Another option is to use an electronic flash, which can have a 1/100,000s – 1/50,000s duration. Are either of these realistic options for a baseball game? Probably not. Flashes are ineffective at long ranges with large amounts of foreground clutter (aka seats and people) and a 1/50,000s exposure with available light is almost certainly going to be too dark to be usable without excessive noise. To go any further, we’re going to need a crash course in photographic exposure.

If you’re the kind of person who just puts the camera in auto and pushes the button, this will help to explain why you get certain results in certain circumstances. If you already know about exposure, then you can skip ahead. Or you can get yourself a copy of Understanding Exposure and come back when you’ve mastered the concepts. Or you can just grab your camera and mess around with the controls until it starts making sense, that’s what I would do. But for now, let’s cover the basics.

Skip ahead to camera discussion
Skip ahead to conclusions

The main factors that control how bright an image comes out are shutter speed, aperture, and ISO speed. Here’s what you’ll need to know about each of them for this exercise. Note that one “stop” is a factor of two increase (more light) or decrease (less light) in the exposure.

Shutter speed is the amount of time that the shutter is open and allowing light to reach the sensor. It is typically measured in fractions of a second, though longer exposures can be measured in minutes or hours. For shutter speed, one stop equals a factor of two increase or decrease (e.g., 1/500s to 1/250s increases by one stop, 1/500s to 1/1000s decreases by one stop). Increasing the shutter speed (decreasing the amount of time) improves the ability to freeze motion at the expense of limiting the amount of light.

Aperture is the size of the opening through which light passes to reach the sensor. It is measured in f-numbers (f-stops), which are typically expressed as denominators relative to focal length (e.g., f/2.8). For aperture, one stop is a factor of the square root of two (~1.4) increase or decrease (e.g., f/2.8 to f/2 increases by one stop, f/2.8 to f/4 decreases by one stop). Increasing the aperture (decreasing the f-number) increases the amount of light at the expense of limiting the depth of field in the focal plane (wider aperture equals smaller range in focus).

ISO speed is the sensitivity of the sensor. In film, ISO is a function of the chemical structure of the film. In digital sensors, ISO is a function of amplification of the signals from the sensors. For ISO, one stop is a factor of two increase or decrease (e.g., ISO 200 to ISO 400 increases by one stop, ISO 200 to ISO 100 decreases by one stop). Increasing the ISO allows for brighter exposures with less light at the expense of increased noise.

Put it all together and you get a complicated equation for exposure, EV = log2(N^2/t) – log2(ISO/100), where EV is the exposure value relative to a 1s exposure at f/1.0 and ISO 100, N is the f-number, t is the shutter speed, and ISO is the ISO. The result is a relative measure of the amount of light in the scene you’re trying to photograph.

Back to our starting image, we have a shutter speed of 1/1600s, an aperture of f/4, and an ISO of 3200. That gives us an EV of 9.6 for the start of a night game just before sunset. What about the end of a night game?

This is a shot from the final inning of the same game, now well past sunset with only the stadium lighting providing illumination. This time, we have a shutter speed of 1/400s and an aperture of f/2.8 to go with the ISO of 3200, for an EV of 6.6. However, even that doesn’t tell the whole story. In this case, I adjusted the automatic exposure to get a higher shutter speed using exposure compensation. By manually decreasing the exposure by 2/3 of one stop, I traded a darker exposure for a faster shutter speed. This is a useful trick when dealing with spotlights that throw off the camera’s metering or when you want to adjust the shutter speed from aperture priority mode. This means that the actual EV level was about 6, meaning that we lost three and two thirds stops of light over the course of the game, or about 90% of the light. A similar shot from a different stadium yielded an EV of 7 (1/500s, f/2.8, ISO 3200), so your lighting may vary. What about day games?

Here’s Alonzo Harris bunting at 2pm in Manchester, NH, which, considering daylight savings time, is just an hour or so from when the sun was at its highest. The shutter speed of 1/3200s, aperture of f/4, and ISO of 400 give us an EV of 13.6. That’s four stops better than the start of the night game, or about 16 times as much light.

These three examples give us a range of 6 to 13.6 for lighting levels across the range of baseball game conditions, but what does that have to do with freezing a baseball? For that, we need to get into hardware.

There are a lot of choices for cameras and lenses to use at baseball games. While a good photographer can get good photographs of some kind with any camera, that often comes from understanding the limitations of the equipment and working with its strengths. That also means that, to get the best results, you will need the right tool for the job. In this case, that means a full-frame SLR and a long telephoto lens with a wide aperture. I’m most familiar with Canon cameras and lenses, but you can find equivalents for most common brands.

When you see a professional photographer working a game, they’ll usually have at least one camera with a 300mm f/2.8 or 400mm f/2.8 lens. These are the focal lengths it takes to frame the action from the typical distance of the camera well. The maximum aperture of f/2.8 gives the best chance of maintaining a decent shutter speed in low light, though our night game example showed that this could still be as little as 1/400s with ISO 3200. The downsides of these lenses are significant though. They typically weigh between 5 and 10 pounds, which is more than most people can handhold for extended periods of time. That means you’re going to need a monopod or tripod to keep them steady. That plus the sheer size of these lenses (8 to 14 inches) could cause some problems with the people around you or the stadium staff (most stadiums prohibit professional equipment in the seating area, and these lenses are hard to pass off as anything but). And then there’s the cost, which is firmly in the “If you have to ask…” range (though rentals are a reasonable alternative). This is the best case scenario, but it isn’t realistic for most people.

The next step down is the type of setup I use with a zoom lens that goes out to at least 200mm, typically either a 200mm f/2.8 or 200mm f/4. These are light enough to handhold and can still get you close enough to get decent action shots. If there’s enough light, you can add an optical teleconverter to add range at the expense of maximum aperture. With a 1.4x teleconverter, a 200mm f/2.8 lens will act like a 280mm f/4 lens. This puts you at a one-stop disadvantage compared to the pro option, but it makes more sense for most people.

Before you say anything, yes, I know all about your superzoom lens that takes wonderful pictures and is the only lens you’ll ever need. Lenses like the 18-200mm f/3.5-f/5.6 are great for people who only want to use one lens, but that f/5.6 at the long end puts you another stop down compared to the previous option. It also won’t work with a full-frame sensor, but we can use the 70-300mm f/4-f/5.6 as a substitute.

Anything that doesn’t fall somewhere in the above range isn’t really worth mentioning, and we haven’t even touched on image quality. The other end of the equation is the camera and it goes about the same as the lenses.

On the pro end you have full-size full-frame sensor SLRs like the 1D X. These cameras have the highest quality sensors and processing, including high frame rates, extensive video options, and the ability to produce usable images at very high ISOs.

The next step down is the family of mid-sized full-frame SLRs, represented by the 5D series in the Canon line. The combination of a full-frame sensor and high ISO options at a more reasonable price make these the best way to get good low-light performance without spending a fortune (assuming that you don’t consider $5,000 for the camera body and lens to be a fortune). ISO 3200 is reasonable at this level, with the possibility of using higher ISOs as well (with diminishing results).

Below that is an extremely wide range of APS-C sensor cameras. The smaller sensor means that, while the image appears to be framed like it would with a longer focal length (200mm has the field of view of a 320mm lens on a full-frame camera), you’re also working with less than the total amount of light that a full-frame lens provides. Combine this with increased pixel density (most of these cameras have about the same number of pixels as their full-frame counterparts in less than half the space) and you get less impressive low-light performance. ISO 1600 should be usable, but things can get iffy from there.

Once again, we’re looking at about a one-stop difference between each level, or two stops between each level with camera and lens combined. With the middle tier as our reference, we can get back to the numbers and see what performance we can reasonably expect. First off, we’re not going to freeze a baseball from 100 feet with a 20+ megapixel sensor. That’s just not happening. The images on this site are reduced by about a factor of eight in each dimension for in-line versions and a factor of four for linked versions, so that gets us down to a shutter speed of about 1/6400s to get the equivalent of frozen motion in small scale. So what will get us there?

Results

In our day game example, 1/6400s could be achieved at f/5.6 and ISO 1600, f/4 and ISO 800, or f/2.8 and ISO 400. This should be possible with any set of equipment that can reasonably capture baseball action. In our evening example, we can get to 1/6400s with f/2.8 and ISO 6400. Anything else isn’t going to work. It only gets worse from there, as this ISO table shows.

Approximate ISO required for 1/6400s shutter speed at different apertures

EV	2.8	4.0	5.6
13.6	400	800	1600
9.6	6400	12800	25600
7.0	51200	102400	204800
6.0	102400	204800	409600

Converting to shutter speeds assuming ISO 6400 for the f/2.8 column, ISO 3200 for the f/4 column, and ISO 1600 for the f/5.6 column gives us this:

Fastest shutter speed possible for f/2.8-ISO 6400, f/4-ISO 3200, and f/5.6-ISO 1600

EV	2.8	4.0	5.6
13.6	N/A	N/A	1/6400s
9.6	1/6400s	1/1600s	1/400s
7.0	1/1000s	1/250s	1/60s
6.0	1/500s	1/125s	1/30s

Conclusion

That’s not a pretty sight. From this, we can conclude that you’ll need sunlight in order to freeze the motion of a baseball. The problem with that is that sunlight can cause glare and harsh shadows. Ideally, you’ll want less direct light, either with clouds or wide shadows that cover large sections of the field. Once you get to that point though, it’s a race against time until there isn’t enough light for fast action. And that’s all assuming that you’re in a good position near where the camera wells are (or would be) located. Increase the distance and it might not matter anymore because you could have a hard time even seeing the ball.