Q Okay, I keep hearing about rolling shutter this and rolling shutter that, and how bad it is. You shouldn’t be using this camera because it’s CMOS and you’re not going to be happy with the results. I’m a little confused because, if it’s so bad, why are people still using these cameras? It seems to me that if they were that bad, people wouldn’t be using them.
A Rolling shutter does seem to be a popular topic, as yours isn’t the only e-mail I’ve received on the subject. It seems to create a lot of uncertainty or, in some cases, certainty that’s misplaced—as in “You shouldn’t be using that camera!” Rolling shutter has become popular to discuss since the advent of HD-DSLR cameras but, as I’ll talk about later, it also can be an issue for any CMOS-based camera.
What is rolling shutter? Cameras that use a complementary metal-oxide-semiconductor, or CMOS, chip are unable to capture the entire scene in a single instant. Instead, the image data is captured line by line as if a mechanical shutter was rolling through the device. (As an aside, while it’s common to think that the scanning starts at the top of the chip and travels down, the image hitting the sensor actually is inverted. This is how the lens presents the scene to the image sensor surface. Since the image is upside down, the scanning starts from the bottom of the chip, which is actually the top of the image.)
Using a rolling shutter means that movement in the image—either by subjects moving or camera movement (or both)—exhibits some strange artifacts or geometric distortion. Rolling shutter also can cause problems with bright flashes of light, such as when a camera flashes or lightning appears in the scene. Furthermore, it’s interesting to note that these types of artifacts aren’t something new. Think back to interlace recording and display. Standard-definition television cameras that used tubes rather than chips to capture video didn’t capture the images all in one shot. An interlace tube camera scanned every other line, one line at a time. But we didn’t notice these artifacts. The display, a CRT, was displaying the image using the same scanning method, so the artifacts weren’t apparent. They became apparent anytime you froze the image, by either pausing a videotape machine or grabbing a freeze frame using a frame store or digital video effects (DVE) machine. The frozen image would show slanted buildings and trees when the camera was panned.
It stands to reason that if we shoot with a CMOS camera and are concerned with rolling-shutter artifacts, if we could match the scanning style and speed of the display device with the scanning style and speed of the capture device, the problem would disappear in normal scene playback. Unfortunately, trying to match the display scanning to the camera is an unrealistic and impractical goal, particularly since the scanning speed is very much camera-dependent.
Back to the artifacts themselves. In addition to possible level changes when quick bursts of light are in your scene, I also mentioned that there’s geometric distortion. This is the most noticeable problem. The distortion can be “shear” where objects or edges appear to be slanted. It also can be “wobble” where the objects are compressed vertically as the camera or object moves through the scene.
These distortions aren’t uniform: They depend on how fast the movement (of the object, the camera, or both) occurs, and they vary depending on the camera model. In short, they’re incredibly complex.
So rolling-shutter artifacts are something to be aware of, but like any other technology that you use in production, you need to consider both their weaknesses and their strengths. If you really were guided just by what the weaknesses are in the available production gear, I think that soon you’d think about changing mediums.
For example, I could tell you that lighting instruments give off a lot of heat. Would you stop lighting scenes and only use natural or ambient light? No, you’d recognize when heat would be a problem and choose instruments or techniques that reduce heat. Or, if the first thing you heard about camera dollies is that they’re heavy, would you stop moving the camera? Harsh examples, I know, but I think it’s good to consider how we choose the tools we use rather than just placing gear in the “avoid” category.