The 4K First Mile: Launching 4K Streams

For the purposes of this article, I am going to be discussing the UHD 4K 3840x2160 video resolution, otherwise known as Ultra High Definition (UHD-1). This specific resolution has 2x the vertical and 2x the horizontal resolution of the 1920x1080 HD standard and 3x the vertical and 3x the horizontal resolution of the 1280x720 HD standard. All three resolutions have 16:9 or 1.78:1 aspect ratios. DCI 4K, on the other hand, has a slightly higher horizontal resolution of 4096x2160 pixels and a 256:135 or 1.90:1 aspect ratio.

4K Camera Selection

UHD is the standard that has widely been adopted for televisions, computer monitors, and 4K video recordings, whereas DCI 4K is the cinematic standard intended for movie production and big screen movie theater playback. As such, DCI 4K recordings or output is not standard on most video cameras as it is a higher-resolution standard than what a UHD video camera can record or output.

Fortunately for video producers who want to produce a 4K live stream or webcast, UHD is the 4K standard that you most likely will be working with. UHD 4K workflows are much less expensive to work with for the simple reason that UHD equipment is less expensive than DCI 4K equipment, especially with video cameras and video capture cards.

Selecting a 4K video camera for a UHD 4K webcast workflow doesn’t have to be an expensive or complicated process. What makes webcasting in 4K (and in HD, for that matter) very different from producing content for on-demand viewing is that a webcast is uploaded live. Fast and consistent internet upload speeds are not always available, so webcast producers tend to be more aggressive in selecting a webcast bitrate. Too high a bitrate, and you risk dropping frames on the upload due to lack of internet upload speed. Or too slow a ping time and too low a bitrate, and you run into a situation known as pixel starvation, where you don’t have enough data to reproduce scenes with transitions or lots of action. Pixel starvation results in macro-blocking, and many other undesirable artifacts that can all be grouped together with the result being a poor-quality video stream. Ultimately, if you don’t have enough bandwidth, it is better to lower the video resolution than to push a poor-quality, pixel-starved video stream.

Because bitrates for webcasts are more aggressive than for on-demand viewing, the best thing you can do for your video quality is to start with a clean video signal. Audio engineers always tell you that a strong signal-to-noise ratio is key to great audio and that microphone placement is more important than having a great microphone that is not near enough to the audio source. The same principle applies to video, except the key to a strong video signal has less to do with placing the video camera right in your subject’s face and more to do with making sure that you have enough light on your subject so that you do not require a high ISO or gain to digitally brighten your image.

Outdoor daytime locations typically have enough light, although controlling the difference between the highlights and the shadows is another challenge, especially if your subject moves between lit and shady areas, requiring you to change exposure constantly. I’m not going to go deep into this workflow, but modern 4K cameras, with the ability to film with LOG curves and to apply LUTs into a live workflow, go a long way into managing this challenging lighting situation. Working with LUTs in a live workflow can be challenging, so I am looking forward to working with the new Hybrid Log-Gamma standard, that was jointly developed by the BBC and NHK, and can be viewed on SD monitors and supported HDR displays without any color grading or application of a LUT.

My point about lighting is that you need enough light to reach your video sensor to avoid having to use gain. Higher levels of gain introduce noise into your video signal, and noise is one of the biggest challenges for long GOP interframe codecs. This is because noise is random in nature and steals valuable bits that could be better used to improve the video quality of your subject instead of being wasted trying to reproduce noise.

Lower-noise video cameras are typically video cameras that are better in low light. The feature on a video camera that is the best predictor of low-light sensitivity is the size of the image sensor. This comes down to basic physics, and when I am trying to resolve a 4K UHD video signal, sensor size is what I look at first. It doesn’t help that video sensors have different units of measurement and aspect ratios. As a Canadian, I work with metric and imperial measurements interchangeably all the time. But while this works well enough for common weights, heights, speeds, and distances, it is more challenging to compare tiny measurements in the fractions of inches with terms like APS-C, Super 35, and 35mm full frame. The sensor measurement that I like to use as a basis for sensor-size comparison is the diagonal measurement. This measurement can vary depending on the aspect ratio of the sensor, but as a rough guideline, here are some common sensor sizes and their diagonal measurements:

1/3" = 6mm
1/2" = 8mm
2/3" = 11mm
4/3" = 20.4mm
APS-C = 26.7mm
Super 35 = 28.4mm
35mm Full Frame = 43.3mm

If you live in the U.S., then millimeters might not mean much to you, so I want you to focus on the difference in size between these sensor sizes, regardless of the unit of measurement. If we pretend for a moment that these diagonal measurements are inches on a monitor, then the 1/3" sensor, scaled up many times, is about the size of an on-camera video monitor, and the 35mm Full Frame sensor, scaled up the same amount, becomes a huge 43" 4K UHD computer monitor. Asking a 1/3" sensor to record a decent UHD 4K signal is like asking you to use a 7" on-camera monitor for word processing. I’m sure you could do it, but it would cause you a lot of eyestrain and leave a lot to be desired.

You don’t have to film everything with a 35mm full-frame sensor to attain professional 4K UHD results, but if you follow video camera sensor size evolution like I do, you will have noticed that there has been an increase in camcorders with 1/2" sensors, and that the newest sensor size is a 1” sensor. I am a big fan of the 1" Sony PXW-X70 and PXW-Z150 camcorders (Figure 1, below), because in a camcorder form factor, they have a cleaner video signal in lower-light situations, and the larger sensor helps deliver a sharper image with a shallower depth of field than a camcorder with a smaller sensor.

Figure 1. The Sony PXW-Z150 1" sensor 4K camcorder