In 2006, Sportvision implemented their first camera system for baseball analysis, PITCHf/x. Two cameras are set up in every stadium to record speed, movement, and location of every pitch. While there are occasional flaws, it has become an extremely reliable source of evaluating a pitcher's "stuff." The common values listed under PITCHf/x pages are pitch type percentage, average velocity, horizontal movement, and vertical movement. The first two are self-explanatory for the most part, so I'll explain how the movement works.
The movement values are a comparison in location to a ball thrown without spin from the same trajectory. For vertical movement, negative is downward break and positive is upwards, as one would expect. For horizontal movement, negative is in towards RH hitters and positive is towards LH hitters, so essentially the catcher's view. Texasleaguers.com has a great tool to filter out certain time periods and get league averages for each type of pitch.
Some misconceptions about pitch types can be pointed out with this tool. On average, 4-seam fastballs still have about 5 inches of tail and sliders only "slide" about 2 inches laterally. Cutters are very interesting, showing nearly no horizontal movement on average. Guys like Rivera do have some glove-side movement, but the main illusion comes from the fact that the ball isn't tailing like other fastballs.
There are a couple problems that require some attention. Sometimes pitches are classified incorrectly or separated/combined wrong. For instance, Scott Baker's main fastball is classified as a sinker. Pitchers who throw 58% sinkers don't have a ground ball rate around 35%. A couple common pairs of pitches to be incorrectly separated are sliders/cutters and changeups/splitters. The cameras are not always calibrated correctly at a certain stadium, creating biased results. One such example was at Kaufmann Stadium in 2010, where pitches were reading about 1 MPH faster than other stadiums.
As interesting as PITCHf/x is, Sportvision is working on three more systems that are exponentially better: HITf/x, FIELDf/x, and COMMANDf/x. HITf/x calculates speed off bat, angle of trajectory, and contact point. This should be able to show if a hitter has the same pop or if a pitcher is causing weaker contact or just getting lucky. FIELDf/x calculates reaction time, route, and player speed. These raw numbers would provide much more credibility to defensive metrics. COMMANDf/x shows how much pitchers miss their catcher's target. This will show the quality of a pitcher's strikes and, to an extent, the ability of a catcher to call a game. These are extremely exciting endeavors and they could be the root of true player evaluation in the future.