- Pro Video
- Lighting & Studio
- Pro Audio
- TVs & Entertainment
- A/V Presentation
- Shop Categories
- Used Dept
This chapter discusses the many benefits that HDSLRs have to offer, as well as the limitations that have to be considered when making the choice of adopting these cameras. Each benefit or limitation needs to be weighed against the specific needs of each user. However, there will always be situations where HDSLRs may not be the most effective choice. Professionals who rely on their camera may not be as flexible as an indie filmmaker who may be more willing to work around HDSLR limitations.
Additionally, not all HDSLRs are created equal. Some carry more benefits in certain areas than others. Only your particular requirements for a project or personal shooting style will determine the best choice of camera. The topics in this chapter seek to clarify each determining factor for selecting the appropriate HDSLR for your needs.
The size of the sensor has a profound effect on the resulting image. Generally speaking, the larger sensors will be better in low light (with less noise), will have more dynamic range, more relative shallow depth of field and have a wider field of view. Each of these characteristics is generally desirable, especially for video. On the other hand, smaller APS-C size sensors closely match the size of 35mm film for cinema, making it easier to accurately focus with more depth of field.
Ideally, the best way to record video would be to record the full information from the sensor—otherwise known as RAW. In reality, only high-end and expensive camera systems offer this option. In order to record a manageable amount of information, the video must be compressed using a Codec.
Codec is short for coder-decoder—which as its name implies—encodes the information from the sensor in a compressed format to be later decoded by a compatible player or software. HDSLRs use different codecs for recording. The three main formats are H.264 for Canon, Motion JPEG for Nikon, and AVCHD plus Motion JPEG options for Panasonic.
While excellent for Web delivery, these highly compressed formats were not developed for professional video recording applications. In brightly lit situations, there will be hardly noticeable issues, but when working in low light situations, the codec can sometimes show its weakness. Therefore it's important to understand the camera's limitations.
All Nikon HDSLRs, except for the D3100, utilize the Motion JPEG, or MJPEG codec. Panasonic offers an MJPEG option on the GH2 as well. MJPEG is an intra-frame codec; it compresses each frame as a discrete unit, without reference to any other frame. This makes it easier to edit, encode and decode, but it lacks the sophistication of codecs such as H.264, which take advantage of the similarity between adjacent frames to achieve better compression. Theoretically, more bitrate is required to produce MJPEG footage with quality comparable to H.264. MJPEG is currently more widely supported by editing software than H.264.
All Canon HDSLRs use the H.264 codec while the Pansonic offers an AVCHD option, which is a subset of H.264, applying a limited selection of the compression techniques specified in the full H.264 standard. Because of the H.264's predictive compression, fast-moving objects or panning across a frame quickly can produce artifacts in the image. Either technique is theoretically capable of better results for the bitrate than MJPEG.
Bitrate is the file size of the video over a period of time, usually specified in megabits per second (where eight megabits is equal to one megabyte on a hard drive). The bitrate can be used to judge the amount of compression being applied to a video (smaller rates mean the video is being compressed more). However, more advanced compression schemes such as H.264 and its derivatives may achieve better picture quality per unit bitrate than simpler types such as Motion JPEG. Add to that the fact that some cameras implement the H.264 codec better than others with the same codec (and still be entirely within the specification), and it's clear that bitrate alone can't be the sole determining factor of quality.
Three things control color accuracy in digital video: colorspace, subsampling and compression. Compression is a familiar concept that produces artifacts that most people are familiar with. Subsampling is more complex; it's effectively a video compression technique, dating from the dawn of color television, which takes advantage of the fact that human vision is far less sensitive to color than to brightness. YUV video formats, including most kinds of H.264, store color images as a full-resolution black and white image representing brightness, and a pair of images at reduced resolution representing the color offset. The familiar ratio notation represents this: 4:2:2 means 4 luminance (Y) samples for every two color (U/V) samples, with the color resolution half of the luminance in the horizontal direction. Confusingly, 4:2:0 doesn't mean 4 Y samples, 2 U samples and no V samples; that would produce a very odd-looking picture. It’s shorthand for saying “4:2:0 on one row, followed by 4:0:2 on the next”—the U/V channels are reduced in resolution both horizontally and vertically, which is what most HDSLRs do.
Colorspace itself is a simple idea with far-reaching consequences. Color images are always made up of three values because we have tristimulus eyes, and the idea of colorspace is valid for any given set of those values. The easiest way to think about it, though, is in terms of red, green and blue: specifying a colorspace tells us exactly which red, which green and which blue we're talking about. This controls the range of color we're capable of representing. Most HDSLRs will shoot in a colorspace known as sRGB, the colorspace to which computer monitors are built to conform.
ISO (a common acronym for the International Organisation for Standardization) is a system originally developed for photochemical film, representing the relationship between exposure and the density (or darkness) of the resulting photographic plate. In simple terms, ISO is a standard used to measure the light sensitivity of the sensor. The higher the ISO, the more sensitive the camera becomes to light. All electronic sensors must simulate ISO based on their native sensitivity, so changing the ISO artificially boosts the signal coming from the sensor. Because every sensor has an inherent amount of noise to begin with, boosting the sensitivity of the sensor also boosts noise in the image. In video terms, this is called "gain." However, the difference with HDSLRs is that their larger sensors allow for more light-gathering capabilities and lower noise floors, so the "gain" can be increased more than that of video cameras without causing as much noise.
The full-frame and professional lines of HDSLRs have the lowest noise floors and can therefore push the signal more than lower-end HDSLRs.
For the most part, sticking to an ISO of under 800-1600 will produce a reasonably clean image and with newer HDSLRs such as the Nikon D3s and Canon 1D Mark IV, the limit can be pushed even further and still result in a clean image.
The full resolution of a DSLR's sensor is enormously larger than HD, but unfortunately, the cameras don't base the HD frame on that original sensor data. Instead, most cameras reduce resolution by skipping pixels on the image sensor, which creates aliasing and artifacts, called moiré, that are especially apparent on finely detailed subjects (and are almost impossible to remove). Because of this issue, the effective resolution of the footage will be less than the ideal resolution of 1080p or 720p. However, not all HDSLRs experience this problem to the same degree.
Shallower depth of field means that a background can be blurred beyond recognition, which can have a pleasing aesthetic and can be used to focus the audience's attention on foreground parts of a scene to tell the story. Full-frame sensors can help achieve a shallower depth of field than smaller sensors—mere fractions of an inch in some cases.
While this can sometimes look nice, the resulting image can also look over-blurred. Pulling focus with a moving camera/subject becomes nearly impossible, resulting in that "seeking focus" look. One of the benefits of shooting with an APS-C size sensor, like the Canon 7D, is that the videographer can open up the aperture (to shoot in lower light), but still have a manageable depth-of-field. On the other hand, with a full frame camera, the aperture must be stopped down further to keep the depth of field in check, which means more light will be needed.
The field of view, or FOV, of full-frame sensor cameras is naturally much wider than that of video cameras with a lens of the same focal length. This allows the camera operator a choice of lenses, because a wider field of view can be achieved without the use of super-wide lenses (which have their own limiting characteristics). In order to achieve the same FOV as a full frame sensor on the smaller APS-C sensor, a focal length of 1.5 or 1.6 times less is needed. This means having to use more expensive, wider lenses to achieve the same FOV. On the other hand, there is a benefit that cropped sensors have for shooters who need a higher "zoom" factor. Longer lenses cost a lot more, so being able to use a short lens that effectively acts like a longer lens can be beneficial.
It is worth noting, however, that other than the field of view, lenses do not change their optical characteristics (e.g. barrel distortion) when placed on a cropped sensor, and although the crop may exclude problems that exist only toward the edges of the lens, glass that suffers these problems may have other issues as well.
Working with different frame rates allows for more flexibility in either delivering footage that meets a certain standard or for having more creative control. The standard for film is 24 fps, which tends to give the video a more cinematic look because of its similarity to the frame rates of motion pictures. Faster frame rates enable slow-motion shots without having to duplicate frames in post production (which results in choppy motion).
Currently, HDSLR models like the 7D, 1D Mark IV, T2i, and GH2 offer frame rates of 24-, 25-, 30-, 50- and 60 fps. It’s worth noting, however, that the higher frame rates of 50 and 60 fps drop the resolution to 720p or less, making it possible for the camera's processor to record at higher rates.
For photographers who have an existing set of lenses, getting into video is that much easier with the HDSLR. Filmmakers will also be delighted to know that they can choose from a vast assortment of lenses to fit any situations or budget—in stark contrast to conventional video-camera lenses.
If the HDSLR's branded lenses aren't enough, or if you have an existing set of lenses from another camera brand, there are many options for adapting lenses that were designed for different mounts. The availability of adapters generally levels the playing field for most HDSLRs. However, there are differences when it comes to mounting higher-end PL lenses.
The advent of the HDSLR has spurred a large interest in many different kinds of still lenses. This is due in part to the fact that autofocus on most HDSLRs—excluding the Panasonic GH2 and a few others, is absent or almost unusable, forcing users to seek out lenses that are better at manual focus. In this respect, cinema lenses are the best option, and now they can be mounted on certain camera bodies.
The rear element of PL lenses extends farther into the camera than that of still lenses, and would otherwise not fit because of the mirror in front of the sensor. The best cameras for mounting PL lenses are currently the Panasonic GH2 and Canon 7D. For sensors larger than APS-C, there are very few cinema lenses that cover the imaging area, so there aren't many professionals modifying their large-sensor HDSLRs to accept PL lenses.
The low-light capability of HDSLRs, when compared with most video cameras, is what helps make the HDSLR unique. Shooting with available light (indoors or night exteriors), or shooting with a smaller lighting package opens up a world of shooting possibilities that once were difficult or impossible to achieve. The sensor size has a significant effect on the low-light performance of HDSLRs, with full-frame sensors taking the lead.
Today’s HDSLR sensors have six to ten times more raw pixels than what is required for HD video. HDSLRs compromise the image quality and resolution in order to capture a much lower resolution and higher frame rate. The optimal approach would be to downscale the full resolution from the sensor to an HD (1920 x 1080 or 1280 x 720) frame, but this, unfortunately, is not how it’s done.
Instead, the camera's processor doesn't read the sensor's full information. Effectively termed "line skipping," the camera skips pixels to make it possible for the camera to process the information. This creates image artifacts called moiré, caused by the irregular interference between detailed patterns in the subject and the pattern of active pixels on the sensor. Certain detailed patterns, such as those on brick walls, rooftops, grids, patterned fabrics, etc., will have red and green/blue bands around the details. As the camera moves, the artifacts appear to move around on the pattern, which distracts the viewers' eyes.
In reality, many video cameras also have to contend with this issue, although only single-chip designs will suffer from chroma aliasing. The solution is an optical low-pass filter—effectively a blurring filter that removes any detail from the image that would cause aliasing. DSLRs have anti-aliasing filters as well, but they are designed only to control aliasing for stills, not video, where the effective resolution of the sensor is much lower and would require a much more heavily filtered image. This issue is one of the more serious defects of the HDSLR camera. It should be approached with caution, and shots should always be examined for this problem because it can creep up unexpectedly. Make sure to have a large, high-resolution monitor so that moiré issues can be spotted as they're being shot. This will provide a better chance of correcting the issue—although not all HDSLRs have a full-resolution HDMI output, which can make it hard to spot.
Because of the way that some CMOS sensors are designed, the camera reads the image from the sensor going from top to bottom, rather than all at once. This is called "rolling shutter." The reading process happens very quickly, so if the camera and subject are stationary, there won't be any noticeable effect. Problems begin when a vertical subject or object quickly moves horizontally across the frame—for instance, a bus driving down the street, or a building "moving" across the frame while the camera is panned. Straight lines will appear to lean left or right, depending on the direction of the motion. The faster the object moves, the more noticeable the issue. Aside from being distracting, this issue can make certain post effects, like 3D tracking, much more difficult.
For the most part, the best way to solve the problem is to avoid it. This can be a challenge for unplanned shots, but keeping to very slow pans and panning with a moving object minimizes this effect.
Some HDSLRs offer limited manual exposure control in video mode. For anything more than a quick and mostly preset shot, these cameras prove to be less desirable for most professional applications, especially when there are other HDSLRs that offer full manual control in video mode. In automatic exposure modes, the camera constantly adjusts the exposure at the slightest change in brightenss within the scene.
At present, all HDSLRs have a time limit per video clip (Aside from the Panasonic GH2, which with the exception of the European model, can record until the card is full). The Canon HDSLRs can shoot up to twelve minutes of footage per take, while the Nikons have a five-minute limit.
This limitation is mainly due to the file system used by the cameras (FAT32), which restricts the file size. For anyone who has shot with film cameras, a twelve-minute time limit is completely acceptable. For others, the key, as with the rest of the HDSLR's limitations, is prioritizing what's most important. Having to pause an interview because the clip runs out can be challenging, but many users learn to work beyond this in view of the cameras’ other benefits.
All HD video-capable DSLRs offer an HDMI output. However, some only output a signal when the video is played back – so no monitoring is available before or during recording. As for the rest of the HDSLR lineup, the differences lie in the resolution the camera outputs in different modes. The Canon 5D Mark II, for example, down-converts the 1080i video signal to 480p when it begins recording, but outputs 1080i in standby mode. This limitation needs to be weighed against how the camera will be used. If, for example, an external monitor is needed during recording to assess focus, a 480p-resolution picture will make it difficult to accurately achieve it.
It's worth repeating that HDSLRs were not designed for professional video applications. In stark contrast to the beautiful images that are possible with HDSLRs, the audio quality is impeded by several factors. Low quality built-in mic, low-quality preamp, lack of a mic input (on some models), and no way to disable the horribly hissy AGC (Automatic Gain Control; the Nikon D300s and D3s can limit AGC sensitivity, while the updated 5D firmware offers an AGC disable feature).
For many professionals, the lack of audio capability isn't a deal breaker because most high-end productions use a dedicated audio recorder and mixer for sound. This method of recording can achieve a higher level of sound quality and offers other conveniences, such as dedicated audio controls and the ability to have a recording specialist move around untethered from the camera. In most situations, this is preferred in order to get the mic as close to the subject as possible.