The first video-capable DSLRs were aimed at photojournalists, but were quickly adopted by wedding photographers and even serious filmmakers. The large sensors (relative to those found in pro camcorders) allow for cinematic limited depth of field and thus effective selective-focus shots. HDSLRs are much smaller and far less costly than the pro digital camcorders, and the lens selection is extensive. Plus, you also got an outstanding, high-tech still photo machine. But there’s more to shooting videos with your HDSLR than just pressing the "movie" button. The artistic aspects of filmmaking are beyond the scope of this article, but here are some things you should know about using HDSLRs to shoot videos.
There are a number of video resolutions, and most of today’s DSLRs can shoot several. HD (high-definition) video is 1280×720, meaning each image consists of 720 horizontal lines, each 1280 pixels wide. Full HD is 1920×1080, or 1080 horizontal lines, each 1920 pixels wide. (Note that 1920×1080 is just 2.07 megapixels per frame.) Some DSLRs (and many older compact digital cameras) also can shoot VGA video (640×480, not useful for professional purposes). SD (standard-definition) digital video is 720×480; no HDSLR shoots that.
Full HD 1920×1080 is great, but there are even higher-resolution formats, most notably 4K, with images about 4000 pixels wide. The 4K Ultra HDTV standard is 3840×2160—four times the spatial resolution of 1920×1080 full HD. Canon’s EOS-1D C does 4K at 4096×2160. RED offers still + motion cameras that can do 4K at 3840×2160 (1.78:1 aspect ratio) and 4096×2304 (1.9:1 aspect ratio). RED also offers cameras that can shoot 5K video—5120×2700 (1.9:1 aspect ratio) and 5K WS 5120×2134 (2.4:1 wide-screen aspect ratio). Ultra HD 8K is 7680×4320—16X full HD’s 1920×1080. Announced at NAB in April, RED’s newest camera model, the RED EPIC DRAGON, offers 9X the resolution of HD files with a 6K sensor capable of 6144×3160 resolution with up to 6K and 5K acquisition formats; 2:1, 2.4:1 and anamorphic 2:1 aspect ratios are available, as well as 16:9 when capturing in 4K, 3K or 2K/1080p/720p.
The aforementioned figures are in-camera recording figures, of course. The resulting videos can be played back on a variety of media, from digital cinema (movie theaters) and home HDTVs to computer monitors, iPads and even mobile phones. Today’s digital theaters generally use 2K (2048×1080) at 24 or 48 fps, or 4K (4096×2160) at 24 fps. Standard digital television is 640×480 (4:3 aspect ratio) or 720×480 (16:9 aspect ratio). Apple’s top MacBook Pro has a 15.4-inch Retina Display that uses 2880×1800 (220 pixels per inch). The top iPad has a 9.7-inch Retina Display using 2048×1536 (264 ppi). The iPhone 5 has a 4-inch Retina Display using 1136×640 (326 ppi).
The 4K, 5K and 8K formats are the wave of the future, and if budget allows, it’s not a bad idea to shoot 4K to provide some future-proofing. However, today many pro projects are done quite satisfactorily in 1080p. And keep in mind that the higher the resolution, the larger the image files, and the more powerful the processors needed to deliver usable frame rates in-camera (and afterward, to edit the resulting footage in your computer). Four minutes of compressed 1080 full HD results in a file size of around one gigabyte; 4K files are 4X that size.
Most DSLRs shoot 1080 full HD at 24 or 30 fps (some offer both, and a few even 60 fps). Common 720 HD frame rates also include 24, 30 and 60 fps. VGA usually goes at 30 fps. These are NTSC frame rates, used in the U.S. and some other areas. Much of the rest of the world uses PAL, which has standard rates of 25 fps and 50 fps, something to beware of if you’re thinking about buying a gray-market HDSLR.
Motion pictures on film have long been shot at 24 fps, so the 24 fps video rate gives what many consider "film-like" motion. The faster 60 fps rate produces a more "real" feel to some viewers. The standard 30 fps rate works well for general use. Experiment with the different rates your HDSLR provides and see which you prefer.
With some early HDSLRs, the frame rates really were 24 fps, 30 fps and 60 fps. But the "real" NTSC frame rates are 23.976, 29.97 and 59.94, and most of today’s DSLRs shoot at those rates rather than 24, 30 and 60 fps. (Your DSLR’s instruction manual will tell you which it uses; from a shooting standpoint, there’s no difference between 23.976 and 24 fps.) Where did those odd numbers come from? In the days of black-and-white TV, the broadcast frame rate was 30 fps, half the 60 Hz rate of U.S. AC power. When color TV came along, there was interference between the color signal and the sound, so the powers that be decided to reduce the rate to 30/1.001 = 29.97 fps to eliminate that problem. The 23.976 rate is 24/1.001, and 59.94 is 60/1.001.
Note that you don’t necessarily have to play back videos at the same frame rate at which they were captured. If you shoot at 60 fps and play back at 24, you’ll get a slow-motion effect. Conversely, if you shoot at 24 fps and play back at 60, you’ll get a speeded-up effect. Some compact digital cameras can shoot lower-resolution video at 300 fps, even 1200 fps, which is good for special effects when high resolution isn’t required.
Progressive Vs. Interlaced
You’ll see a lower-case letter "p" or "i" following the resolution or frame rate in video specs: 1920x1080p 24 or 1920×1080/24p. The "p" means progressive; each video "frame" is drawn line by line from top to bottom at the specified rate: 24 times a second for 24p, 60 times a second for 60p. The "i" means interlaced, with the even-numbered lines of the frame drawn first, followed by the odd-numbered lines, so the entire image doesn’t appear simultaneously; rather, you get half the image quickly followed by the other half. The significance here is that interlaced can produce odd artifacts with moving subjects. (Standard U.S. television is broadcast at 60i.)
Frame rate is the number of video frames per second. Shutter speed is the amount of time each of those frames is exposed. As with still photography, the shutter speed controls how sharp a moving subject will appear in each frame. However, there’s a relationship between shutter speed and frame rate. In the old film days, this was known as the shutter angle, as the shutter was a disk whose opening size (angle) controlled the exposure.
F-stops Vs. T-stops
Still camera lenses have F-stops. Those are calculated, not measured: F/4 means the effective opening is 1/4 the focal length of the lens. But production tolerances, light loss due to multiple elements and imperfect coatings and the like result in less light reaching the imag
e sensor than the calculated F-stop would indicate. So movie lenses are calibrated in T-stops (true stops), which are actually measured for the lens in question, and thus accurate indicators of light actually transmitted to the sensor. For some lenses, the difference between the calculated F-stop and the actual T-stop can be significant—one reason why pro movie lenses cost more than typical still-camera lenses. The movie lenses also have manual-focusing rings that are geared "slower" for easier and more accurate manual follow-focusing while shooting motion.
Showing Your Work
HD video files are huge, making getting examples of your work out to potential clients a challenge. Probably the best way to do it is to post the video on Vimeo or YouTube, and send the client a link. A Vimeo PRO Membership provides unlimited HD uploading and lets you control who can watch your videos, and the cost is far less than sending out hard drives. For most natural-appearing video, the shutter speed should be close to 1/2X the frame rate: 1⁄60 sec. for 30 fps, 1⁄50 sec. for 24 fps, 1⁄125 sec. for 60 fps, etc. Faster shutter speeds make the images of moving subjects in each individual frame sharper, but result in a jerky effect when the video is viewed. Conversely, slower shutter speeds result in more blur of each moving subject, but the motion appears smoother when you view the video. You can experiment with different shutter speeds, but the 1/2X frame rate rule of thumb is a good place to start.
Keep in mind that shutter speeds used for video shooting are pretty slow compared to those used for still photography. The fastest common video frame rate is 60 fps, calling for a shutter speed of 1⁄125 sec. per the 1/2X rule of thumb. If you intend to pull still frames from your video, faster shutter speeds are necessary; slow shutter speeds will result in blurred moving subjects, something that’s not that noticeable when watching the video, but would be quite evident in a still image. Pros who are savvy to a still + motion workflow slow down the action on set to get sharper frames and use video mode as a 24 fps still camera to get frames that capture the perfect moments in a movement (see "The Beauty Within," DPP, September/October 2012, page 44). Of course, you can always use your DSLR as a still camera to get top-quality still images.
For most natural-appearing video, the shutter speed should be close to 1/2X the frame rate: 1/60 sec. for 30 fps, 1/50 sec. for 24 fps, 1/125 sec. for 60 fps, etc. Faster shutter speeds make the images of moving subjects in each individual frame sharper, but result in a jerky effect when the video is viewed.
The focal-plane shutter in the DSLR is used only for still photographs. For video, the sensor’s own electronic shutter is used (and a good thing, or you’d run through your focal-plane shutter’s 150,000-cycle life in about 45 minutes of 60 fps video).
CCD sensors use a global shutter; they effectively expose all the pixels simultaneously. CMOS sensors use a rolling shutter; they effectively expose the pixels row by row starting at the top. With still photos, this doesn’t matter much. With video, the rolling shutter can adversely affect moving subjects (and stationary ones when the camera is panned across them) since the subject position will be different as each line of the image is recorded.
Most pro digital camcorders use CCD sensors, and thus don’t have the rolling-shutter problems. Most DSLRs use CMOS sensors, and thus do produce rolling-shutter effects. Newer cameras minimize the effects, as does careful shot planning.
Another consideration is line-skipping. A full HD pro camcorder has a sensor of 1920×1080 pixels. An HDSLR has a sensor with many more pixels than that. To get down to 1920×1080, much of the pixel data must be eliminated. Ideally, this would be done by binning—combining data from several pixels into one. But that would require lots of in-camera processing power—reading a 24-megapixel sensor at 30 fps is a lot of data. DSLRs frequently use "shortcuts." One is line-skipping; they just eliminate, say, two of every three lines to get down to 1920×1080 (2 megapixels) from 22, 24 or whatever is the sensor’s actual pixel resolution.
Codecs And Compression
HDSLRs use a variety of formats and compression schemes. The most popular are MPEG-4/H.264 and AVCHD. Canon and Nikon pro DSLRs record in MOV format, using MPEG-4/H.264 compression. Sony’s SLT-A99 records in AVCHD Ver. 2.0 (progressive) format, which permits shooting 1080 full HD at 60p; it can also shoot MP4 video. RED cameras use RED’s own versatile, high-quality REDCODE RAW format. The newer pro HDSLRs also let you record uncompressed video directly to an external recording device via HDMI.
Some cameras let you choose different compression levels or bit rates. Lower compression and higher bit rates make for better image quality, but larger files.
DSLRs were designed for still photography, with the camera held up to your eye. To shoot video, the camera has to be in Live View mode, and that shuts off the DSLR’s eye-level viewfinder (except with Sony SLT cameras; more on those in a bit). So you can’t use the eye-level DSLR finder to shoot video; you have to use the external LCD monitor (or an accessory external monitor). Holding a DSLR at arm’s length to use the external monitor isn’t conducive to comfortable or steady shooting, so there are lots of accessories on the market to make it easier to use DSLRs for serious video work.
Sony’s SLT cameras employ a fixed translucent mirror rather than the typical SLR mirror. In a typical DSLR, when you activate Live View (required for video shooting), the mirror flips up out of the light path so light can reach the image sensor. When this happens, the eye-level viewfinder goes black (because light no longer reaches it), and you lose the camera’s normal phase-detection AF (because light no longer reaches the AF sensor), so much slower contrast-based AF off the image sensor is employed for video. With Sony’s SLT cameras, the fixed semitranslucent mirror transmits most of the light to the image sensor and directs a small portion up to the phase-detection AF sensor. So you get full-time phase-detection AF, which is much better than contrast-based AF for action subjects—even for video. The SLR’s optical eye-level finder is replaced by an eye-level electronic viewfinder, so you also get eye-level viewing for video. That makes the SLT cameras more ergonomic and much better suited for action videos than conventional DSLRs.
Canon’s EOS-1D C cinema DSLR essentially is a Canon EOS-1D X with 4K/24 fps video capability. Canon’s other current full-frame DSLRs (EOS-1D X, EOS 5D Mark III and EOS 6D) offer the same basic video capabilities: 1920x1080p at 30, 25 and 24 fps; 1280x720p at 60 and 50 fps; and 640×480 at 30 and 25 fps. You can choose two compression types: IPB for smaller file sizes or ALL-I for easier editing. You also can choose two types of time coding: Rec Run (runs while shooting) or free Run (runs whether shooting or not). You can use the built-in mono microphone or an optional stereo mic, and the EOS-1D X and EOS 5D Mark III have headphone jacks for easy monitoring of audio. You can use single-shot AF while shooting, but it’s best to focus movies manually. Canon full-frame DSLRs can use all EF (but not EF-S APS-C) lenses, which range from an 8-15mm fisheye and a 14
mm superwide-angle through an 800mm, including macro and manual-focus TS-E tilt-shift optics.
Nikon’s full-frame DSLRs (D4, D800/800E and D600) all offer the same basic video capabilities: 1920x1080p at 30, 25 and 24 fps, and 24 or 12 Mbps bit rates; and 1280x720p at 60, 50, 30 and 25 fps, and 12 or 8 Mbps. Movies are recorded in MOV format with H.264/MPEG-4 AVC compression, or you can record uncompressed video to an external recorder via HDMI. You can use the built-in mono microphone or an optional external stereo mic, and there’s a jack for headphones. Continuous AF is available during shooting. Nikon full-frame DSLRs can use all current and many earlier AF Nikkor lenses. With DX lenses (designed specifically for APS-C sensors), the cameras automatically crop to DX mode; FX (full-frame) lenses range from 14mm through 800mm, including macro and manual-focus PC-E tilt-shift optics.
RED offers unique modular still + motion digital cinema cameras that also deliver still images, use a wide variety of popular cine lenses and feature some remarkable capabilities. The new RED EPIC DRAGON can do 6K video at 100 fps and 4K up to 200 fps, with 16.5+ stops of dynamic range. The RED EPIC Mysterium-X can deliver 5K video at up to 120 fps, with up to 18 stops of dynamic range. The RED SCARLET DRAGON can shoot 5K at up to 60 fps and 4K up to 75 fps. The RED SCARLET-X can shoot 5K at 12 fps and 4K at 30 fps. Mounts are available for Canon, Nikon, Leica M, cinema-standard PL and many other lenses (retaining autofocus and image stabilization with lenses that provide it). RED’s own REDCODE RAW format provides great versatility. Accessories include viewfinders, monitors, power supplies, recorders, handle and more.
The Sony SLT-A99 can shoot 1920×1080 AVCHD video at 60p/28 Mbps, 50p/28 Mbps, 60i/24 Mbps, 60i/17 Mbps, 50i/24 Mbps, 25p/24 Mbps, and 24p/24 and 17 Mbps, plus 1440×1080 MP4 video at 30 fps/12 Mbps. It can record uncompressed 4:2:2 video to an external recording device via HDMI. You can use the eye-level EVF or the tilting/rotating external LCD monitor for composing and focusing. Full-time continuous phase-detection AF is available, when desired. There’s a built-in stereo mic and a headphone jack, and you can add the optional XLR adapter for pro-level stereo sound. The A99 can use all A-mount Sony and legacy Konica Minolta lenses, automatically cropping to APS-C format when a DT (APS-C) lens is used. Current full-frame lenses range from a 16-35mm superwide zoom through a 500mm, including macro optics.
Blackmagic Design 4K And Pocket Cinema Camera
Blackmagic Design offers a couple of interesting new video cameras. The $995 Pocket Cinema Camera features a Super 16-sized sensor and can shoot 1920x1080p video at 23.98, 24, 25, 29.97 and 30 fps, in either Apple ProRes 422 (HQ) or lossless CinemaDNG RAW formats. It uses a Nikon EN-EL20 lithium-ion battery, takes Micro Four Thirds lenses (and many others, via adapters), has a built-in stereo mic, and stores images on SDHC and SDXC media.
The magnesium-alloy body lives up to its pocket billing, measuring 5.0×2.6×1.5 inches and weighing a mere 12.5 ounces. The new Production Camera 4K is a $3,995 pro model that shoots 4K video at 23.98, 24, 25, 29.97 and 30 fps, and 1920×1080 full HD at those rates, plus 50 and 59.94 fps. It features a 3840×2160 Super 35-sized sensor with professional global shutter, and takes Canon EF and Zeiss ZE lenses. Images are saved on removable SSD recorders. A 5-inch touch-screen LCD provides easy viewing and camera-setting. The Production 4K provides focus peaking for easy manual focusing.