When professional photographers began to embrace digital photography, it was, for many, a tentative relationship at best. Digital was seen as being more efficient than film, but the amount of time needed to learn new post-processing workflows seemed onerous, and the amount of gear required to switch over was considerable.
A similar opportunity is opening up for photographers looking to expand beyond photography and to capture video from the same devices as they use for their stills. There are more choices, with higher-quality video and a greater range of outlets for the video captured by someone with a photographer’s uniquely creative eye.
With digital photography, the flexibility of shooting without film opened up new avenues for photographers, with many finding markets and customers they previously couldn’t reach. The early embrace of digital photography allowed smaller studios to compete for jobs with bigger ones, and allowed a new era of shooters who could deliver stunning work without the workflow limitations of waiting for film to come back from the lab.
For the outdoor photographer, this benefit was even greater. Unencumbered by the physical size of film canisters and freed from concern for film’s temperature limitations, outdoor photographers could continue to travel farther, shoot longer and keep working in a greater range of environmental conditions.
The same changes in video are enabling photographers to shoot both personal and client-requested video without having to switch bodies or even lenses. Digital cameras have gotten a huge range of new video features as a result of the manufacturers’ rush to surpass their competition in the video space. Partially the push to video is because the margins between pro cameras is becoming ever slimmer, and so they’re seeking to outdo each other in video, but partially it’s a natural evolution.
Digital cameras, and especially mirrorless ones, are essentially video cameras that happen to capture individual frames instead of a sequence of moving pictures, and digital still cameras are gaining more video functionality as a result. In fact, I’ve been predicting for years a time when companies will produce cameras that essentially capture video, but in still mode extract one (or more) frames from the stream of data coming from the camera. This year a number of announcements from Olympus, Sony and especially Panasonic showed that the future of still being a subset of video is around the corner.
To understand the current market for digital cameras that offer professional-level video, it’s first important to understand what separates digital still and digital video camera functionality.
Still Images Versus Digital Frames
The first digital cameras were an evolution of still cameras. In the very first iterations, engineers simply pulled the backs off of film cameras and made new parts that put a sensor in the path of light instead of film.
Up until mirrorless cameras gained popularity, digital cameras operated very traditionally. Light enters a lens, a shutter opens and lets light hit the sensor, the shutter closes to the stop light, ending the exposure. Rinse, repeat. (This and the following are an oversimplification of the process for clarity’s sake.)
At its heart, a digital imaging sensor is really an analog device. As light hits the sensor, a charge builds up in each pixel, the way it would in a solar panel. A chip inside the camera reads the voltage level in the pixels and converts each into a number. In order to process any more images, the pixels have to be reset back to zero, otherwise more light coming in would just add to the data, making a double exposure.
The closing of a mechanical shutter gives the camera a chance to flush the voltage from the sensor, bringing it back to a fresh state. Video cameras, by comparison, don’t have mechanical shutters, and so they need to use other tricks to reset the sensor back to zero. Many of these techniques skip reading part of the sensor at any given time (referred to as line skipping) to allow video cameras to process images more quickly, but this also reduces the image quality relative to a still image.
Take a still photo with a pro DSLR and take a frame of video (even today’s 4K standard), and you’ll see that a frame of video is much lower quality, generally speaking. The human eye, though, doesn’t need the same quality when looking at a moving image as a still one—the details blur together.
That means that today’s digital cameras, with powerful, high-resolution sensors and fast processors, are more capable of capturing cinema-level video with ease. A 4K video is really a succession of 8MP images captured 30 times a second, so the most powerful still cameras take the image from a larger sensor and make 8MP moving images out of it, rather than only reading 8MP of the sensor. This technique, called oversampling, is what allows the top still cameras to create cinema-level video. It’s analogous to taking a large image in Photoshop and resizing it down for web use, rather than only using a camera that captures at web resolution.
The Power To Move Your Images
While the technology behind still and motion-imaging sensors has been improving, so too have the standards for video. HD video, with its resolution of 1920×1080, has been replaced with UHD (also called Ultra HD, Ultra-High Definition or simply 4K) video, which has 3840×2160 pixels. Cinema 4K has 4096 on the longest side (and a different vertical resolution depending on the aspect ratio.)
That means that even 12MP cameras have the theoretical ability to capture 4K cinema-quality video at full resolution with oversampling, though not all camera manufacturers oversample. Some of the cameras that are capable of shooting 4K video only do so by reading the portion of the sensor equal to the resolution of 4K video. So, a 20MP sensor might crop down to only read 8MP worth of the middle of the sensor. This allows the cameras to process the 4K video extremely quickly, but the tradeoff here is that lenses have an effective focal length modifier just like using a full-frame lens on an APS-C camera would.
This also has the effect of changing the amount of blur possible with a lens. A major draw of video production on a still camera was the ability to use lenses with wide apertures for the background blur. Videos that are created from oversampled video, rather than a cropped sensor, tend to be higher resolution and have better detail.
Imagine a video of an elk set against the background of a ridgeline shot at ƒ/1.8. Now imagine that same image at ƒ/5.6—there would be an entirely different look to the photo. Cameras that only have crop-sensor 4K are not only changing the look of the focal length but the possible background defocus as well.
That’s why some cameras offer a choice between cropped video and full-frame video, which is good, especially for wildlife shooters. A 400mm lens has the equivalent of a 680mm lens when shot on a sensor that has a 1.7x video crop, similar to if you mounted it to an APS-C camera.
Binning And Skipping
We talked about line skipping, which is used both to set a sensor back to zero between captures and also as a way to reduce the amount of data recorded. In terms of overall quality, video that doesn’t line skip is better than video that line skips, because each frame only has half the data in line-skipped video.
A related technique is pixel binning, which groups pixels together as one larger pixel, also to reduce processing overhead. As you’d expect, video from a camera that does pixel binning is lower resolution than a camera that does not, all else being equal.
LOG for more detail
Just as a RAW file contains more information than a low-quality JPEG, video cameras have LOG, which preserves more of the dynamic range possible with the sensor. A video camera that can capture in LOG (and there are a number of different flavors of LOG) produces higher-quality video streams than those that don’t have LOG recording, all else being equal.
Another measure of the quality of a camera’s video capturing—especially when comparing it to another camera in the same brand—is the frame rate of capture. Listed as a frame rate number and then the letter “p,” the higher the frame rate, the more information the camera is processing, and the less artifacts you’ll see from frame to frame.
A specification often seen in video recording is the video’s “pulldown,” which is expressed as a ratio like 4:2:2. Simply put, this is a notation of how color data is sampled. Video is broken into three streams of data, with each number representing one of the three streams. The first number is how many times luminance (brightness) is sampled, and the second two are basically how many times the colors are sampled compared to luminance. In 4:4:4, all three streams are sampled to the same resolution. At 4:2:2, which is TV-quality, each color stream is sampled two times for every four times brightness is sampled.
The gold standard is 4:4:4; 4:2:2 is very good quality and 4:2:0 is not very good looking, generally speaking, compared to the others. A camera that records at 10 bit produces higher-quality video than one sampling at 8 bit (all else being equal), so 10-bit 4:2:2 would have more accuracy than 8-bit 4:2:2 because there’s more color range being recorded from each stream.
With incredibly high data-gathering needs, 4K video often benefits from the use of an external video recorder—a device that can display and record the video stream. Most cameras that can send video out of the HDMI port for recording do so without compression, making the images captured in the recorder higher quality than those captured in-camera.
A final consideration in 4K video shooting is the level and type of compression. Just as a bad JPEG compression can make your images look bad, so too can compression affect the video quality. The compression scheme might not matter when it comes to making a few quick clips of video but becomes noticeable when shooting what is expected to be pro-quality work.
Most cameras use an older compression standard called AVC (also called H.264), which was designed for the delivery of video, not the capture. It’s widely compatible but also doesn’t record all the data a modern video camera can capture. A compatible replacement, H.265, has been released but isn’t widely implemented because of the cost of licensing the technology. Sony has created its own format, XAVC S, which is used both in cameras like the a7S II and the company’s higher-end professional studio video equipment.
The Image Matters In Video, Too
Just as the final image quality is more important than the camera used to capture it, video quality comes down to a number of factors, many of which fight with each other. A camera that captures video with no pixel binning at a high bit rate and with minimal compression problems should, on paper, create better video, but that doesn’t always pan out. So many factors work together in video (sensor, bit depth, pixel binning, compression, lens quality, processing speed, etc.) that it’s often a matter of what the whole package is capable of creating—not individual specs—that should make up a purchase decision.
State Of The Art
With an understanding of the power and the possibilities for photographers to capture stunning video, by far the hardest decision is about what gear to get. If you’ve already heavily invested in a system and own multiple lenses, the best choice may be to pick a video-capable camera in that system, bearing in mind that other systems might offer a larger range of features or the same features in a smaller package.
Within the last year, the video-capable camera market has blossomed, with more systems creating pro-level video than ever, and those features have appeared at much lower price points. Below is a guide to some of the most recently announced cameras that are video capture experts.
The newest models of Canon’s professional cameras, the flagship EOS-1D X Mark II and the prosumer EOS 5D Mark IV, share a lot of the same new video features—and the same limitations in functionality. The systems are close enough in performance to discuss their features together.
Both use the company’s Dual Pixel AF to capture video with continuous tracking autofocus, and both capture Full HD and 4K resolutions. The EOS-1D X Mark II captures HD to 120p and 4K to 60p, while the 5D Mark IV captures 60p Full HD and 30p 4K. The common video distortion called “rolling shutter” is minimized in the EOS-1D X Mark II as well, giving it better image quality than the previous Canon EOS-1D C.
Unlike some competitors, the Canon cameras do not send 4K video across the HDMI port, so there’s no external recording capabilities. Internal recording is 8-bit 4:2:2. Both cameras do cropped video for 4K with a 1.7x magnification factor. Neither offer full-frame 4K video. The compression is unchanged from previous models, with 8-bit AVC/H.264 compression standard.
The Fujifilm X-T2 is not only the newest camera in Fujifilm’s X series, but it’s the first that really takes a stab at the video market.
The X-T2 does cropped video, at only a 1.17x crop—a 35mm becomes a 41mm—and Fujifilm oversamples the video rather than skipping lines or pixel binning. Because of the small size of the camera, a micro-HDMI port is used to send external uncompressed 4:2:2 video at 8 bit. Fuji’s F-LOG format ekes the most range out of the APS-C sensor in the camera. The camera can also capture Full HD at 60p, without any crop.
Fujfilm also eschews the Bayer pattern sensor found in most cameras for its own X-Trans sensor, which it says produces not only better stills but better-looking video as well.
The Nikon D5, like Canon’s flagship, is also capable of shooting 4K video, though it can only do so at 30p, the same as the Canon 5D Mark IV. When it launched, the D5 was limited to capturing three-minute 4K video clips, but a firmware extended that to 30 minutes—sort of. The camera will create this 30 minutes in up to eight separate files up to 4GB in size, and that footage has to be stitched into a single movie in Nikon’s ViewNX-Movie Editor software.
The D5 only does crop 4K video, with a magnification around 1.5x, just a touch bigger than the DX crop would be on an FX body. It supports external recording of 4K over HDMI but does not have a LOG format, and it records in 8-bit H.264.
The APS-C Nikon D500 also captures 4K video and also does cropped sensor video. The focal length multiplier is 1.3x for DX lenses. Since the DX format is already cropped from full-frame, this additional crop is like a crop on top of a crop. The D500 also does 4K UHD at 30p and sends uncompressed 4:2:2 in 8-bit over the HDMI port.
The newest flagship Micro Four Thirds camera, the OM-D E-M1 Mark II, has added 4K video to the top-end model, with the ability to capture 4K at 30p, and can send uncompressed 4:2:2 output across HDMI.
Panasonic beat the competition to 4K video with the 2014 release of the GH4, and it’s looking to create some more firsts with the recently-teased GH5. Announced in September of 2016, the camera won’t arrive until “early” 2017, and all the features haven’t been officially announced. What we do know is that this Micro Four Thirds LUMIX camera will have 4K to 60p and 10-bit 4:2:2 internal recording. We also know it’s recording data at around 6K and sampling down to 4K (as do some other systems) but it will also be the first that allows for “6K” still shots to be pulled from the video stream. When this camera arrives, it’s going to be a compelling video choice.
Sony’s video-oriented a7S II set the standard for still camera video capabilities. While most of the competition created cropped-frame images and/or used pixel binning, the a7S II introduced 4K 30p using the full frame and oversampling. It also has 10-bit video, uncompressed HDMI output, its own S-LOG format and higher-quality compression using its professional studio XAVC S format.
Even though the camera was announced in 2015 (and hasn’t been updated as of press time) the body is still the benchmark for the rest of the 4K-from-a-still-camera market.
This year the company announced the A99 II, which has very similar specs: 4K 30p video, S-LOG 2 and S-LOG 3, XAVC S recording and no pixel binning. The a99 II can capture with a crop or full frame, for flexible shooting. Internal video is limited to 4:2:0 as it was on the a7S II but can send uncompressed 4K at 4:2:2 to an external recorder.
Sony also unveiled the a6500, which does full-frame video and no crop to capture 4K up to 30p, oversampling 1.6x the image data to create a final 4K file. It records in S-LOG2 or S-LOG, and captures 8-bit 4:2:0 video internally and 8-bit 4:2:2 uncompressed over HDMI. The a6300, which was released recently as well, suffered from a bit of rolling shutter; we’re waiting to see if the a6500 has resolved this.