Capturing The Best Video Possible

video primer
The video imaging chain is complex and getting more complex all the time, yet advances in technology are allowing companies like Sony and Panasonic to squeeze immense editing capabilities into tiny still camera-style bodies. These cameras are able to achieve similar features as much more advanced professional video and filmmaking camera systems because they employ heavy compression and a variety of other internal processing tricks that also allow them to capture to consumer-grade CompactFlash and SD media cards. The same compression methods make DSLR and mirrorless systems susceptible to issues in editing and color grading, however.

Even with today’s still camera models, if capturing video internally, there’s a sacrifice to color fidelity and sharpness right off the bat, as the codec—the method of compressing video to fit on a storage device—must be able to squeeze data to write it to CF and SD cards. These cards have fundamental limitations in the amount of data they can process, and so data must be compressed to 8-bit. Compressed 8-bit files utilize chroma subsampling as their principal method of compression. Chroma sampling works by tossing out some chroma (color) to reduce a file and is analogous to JPEG in comparison to a TIFF file.

To express the amount of chroma sampling, companies use a format with three numbers separated by colons, like 4:4:4. While there’s a good deal of color science behind this representation, it simply means that out of 4 bits of information (the first number) how many are used in the final file (the next two numbers). The 4:4:4 is the best as it means all the color information is present. The 4:2:2 skips recording some of the color data, while 4:2:0 skips recording a lot of it. As you’d expect, a 4:2:0 file is much smaller than 4:4:4.

A 4:2:2 is well respected for its ability to keep the weight of video bandwidth down and quality up. You’ll see it often with edit-ready codecs or proxy-editing codecs like ProRes and DNxHD and all of their assorted flavors. Adobe’s Premiere Pro and After Effects, as well as other NLE software, as in nonlinear editing, can work with these file types, so they’re often used for proxy edits or as ready-to-go deliverables for editors.

A 4:2:0 is the quality of compression you find in DSLR and mirrorless systems unless you use an external capture device, but there’s good news. To meet the needs of all these options, video is hungry for disk space. For 4K and high-frame-rate material, newer cards like CFast, the next generation of CompactFlash, have been in use for quite some time, as they’re able to meet sustained levels of data capture without any dropped frames even with 4K. Unlike CF or SD cards, they can also finally record 4:2:2.

Extrapolating further, just like with photography, RAW video is also available, sometimes slightly processed and sometimes unprocessed depending on manufacturer and camera. With any camera, raw video footage must be saved to an external digital recorder through connections like HDMI or HD-SDI—there’s just no way to get this data stored to a CF or SD card fast enough.

A few of the current RAW-capable options include Sony’s 12-bit PXW-FS7 and the Panasonic DC-GH5 with 10-bit output. The Sony a7S II can also output in edit-ready Apple ProRes and Avid DNxHD with 4:2:2 color accuracy and 10-bit color resolution. With the new Sony a9, 4K video is slightly more limited compared to the a7S II and a7R II, as there’s no 10-bit output over HDMI, and the top video bit-rate tops out at 100 Mbits. It still features full-sensor readout with no pixel binning, but professional videographers likely will want to stick to the a7s II. Several dedicated video cameras have internal RAW, as well, including models from Blackmagic Design, which employ Adobe’s open-source CinemaDNG RAW format for up to 12-bit RAW capture.

Before your eyes glaze over with all of these letters and numbers, the number of bits in a video file simply refers to the number of colors capable of being captured. The 8-bit represents a bit-depth with 256 possible variations in color for each of the three RGB channels. That translates to 16.777 million possible colors, or 256x256x256. Stepping up to 1,024 possible colors per channel, 10-bit is a surmountable jump in smooth color gradation with 1.07 billion potential color combinations. The 12-bit rings in with a massive 68 billion, while 14-bits ups the ante to 4 trillion. For practical reference, 10-bit is enough to cover high-dynamic-range video color spaces.

Bit-rate applies to both capture and output, like on a television, for example. There’s a constant battle in video and filmmaking between maintaining consistent input values, i.e., the image information of the frame when it’s captured by the cameraperson, with the output destinations, i.e., the compression and file uniformity that cable, broadcast or Internet channels must utilize to play back to a variety of display devices from iPhones to massive outdoors displays. These include common video playback “container” formats like avi, mov, flv and mp4.

So, while “bit-depth” is the amount of possible color information that can be captured at the sensor, there’s also, confusingly, “bit-rate,” often available through two or three selectable compression schemes. Measured in Mbps (megabits per second), bit-rate refers to the sustained amount of data (the total number of bits) that can be pumped through internal processing every second. The Canon Cinema EOS C300 Mark II has 50 Mbps Long GOP, 35 Mbps Long GOP or 24 Mbps Long GOP, depending on the needs for video file sizes and media capacity.

Bit-rates will also come into play, literally, when streaming to display devices like TVs, computers and mobiles, as the amount of data must stay below the speed of Internet or cellular connections.

There are several codecs available for universal playback, most often called MPEG-4, also known as H.264, the same kind of file you see on DVDs or online. Cam manufacturers have their own unique proprietary blend of internal capture in these codecs. Canon’s XF-AVC employs the MPEG-4 AVC/H.264 video compression format, as does Panasonic’s AVCHD and Sony’s XAVC.

There are also a number of compression techniques that can be used, like Canon’s IPB Interframe, which will use a particular frame to judge movements and contrast to compress image information on something like the background of a scene, for instance, where very little color or contrast will change throughout a scene. The above-mentioned LongGOP, also known as Interframe compression, subtracts the differences from a group of frames to compress image information.

Whether shooting with dedicated video cameras or still cameras, to manage all of this data and to do things like shooting unprocessed, RAW video, HDR video or high-frame-rate video, dedicated video recorders will be necessary for most systems. Many external video recorders these days also come with a monitor with a number of modes for certifying critical picture functions like exposure, focus and dynamic range.

Atomos has become a leader in the field for its top-notch series of monitors and recorders at $500 or less. Its Shogun Blade series is available at only $175. Other companies like AJA, Blackmagic, Cinedeck and Convergent Design also make a number of monitors that can function as recorders.