Lately I’ve done a lot of looking at 4K content. Not only was 4K all over the place at NAB in Las Vegas, but I’ve also had to provide some 4K deliverables on client projects. This has meant a much closer examination of the 4K image than in the past.
First, let’s define 4K. Typically the term 4K applies to either a “cinema” width of 4096 pixels or a broadcast width of 3840 pixels. The latter is also referred to as QuadHD, UltraHD or UHD and is a 2x multiple of the 1920-wide HD standard. For simplicity’s sake, in this article I’m going to be referring to 4K, but will generally mean the UHD version, i.e. 3840 x 2160 pixels, aka 2160p. While 4K (and greater) acquisition for an HD finish has been used for awhile in post, there are already demands for true 4K content. This vanguard is notably led by Netflix and Amazon, however, international distributors are also starting to request 4K masters, if they are available.
In my analysis of the images from various 4K (and higher) camera, it starts to become quite obvious that the 1:1 image in 4K really isn’t all that good. In fact, if you compared a blow-up from HD to 4K of that same image, it becomes very hard to distinguish the blow-up from the true 4K image. Why is that?
When you analyze a native 4K image, you become aware of the deficiencies in the image. These weren’t as obvious when that 4K original was down-sampled to an HD timeline and master. That’s because in the HD timeline you are seeing the benefit of oversampling, which results in a superb HD image. Here are some factors that become more obvious when you view the footage in its original size.
1. Most formats use a high-compression algorithm to squeeze the data into a smaller file size. In some cases compression artifacts start to become visible at the native size.
2. Many DPs like to shoot with vintage or otherwise “lower quality” lenses. This gives the image “character” and, in the words of one cinematographer that I worked with, “takes the curse off of the digital image.” That’s all fine, but again, viewed natively, you start to see the defects in the optics, like chromatic aberration in the corners, coloration of the image, and general softness.
3. Due to the nature of video viewfinders, run-and-gun production methods, and smaller crews, many operators do not nail the critical focus on a shot. That’s not too obvious when you down-convert the image; however, at 100% you notice that focus was on your talent’s ear and not their nose.
The interesting thing to me is that when you take a 4K (or greater) image, down-convert that to HD, and then up-convert it back to 4K, much of the image detail is retained. I’ve especially noticed this when high quality scalers are used for the conversion. For example, even the free version of DaVinci Resolve offers one of the best up-scalers on the market. Secondly, scaling for 1920 x 1080 to 3840 x 2160 is an even 2x multiple, so a) the amount you are zooming in isn’t all that much, and b) even numbered multiples give you better results than fractional values. In addition, Resolve also offers several scaling methods for sharper versus smoother results.
In general, I feel that the most quality is retained when you start with 4K footage rather than HD, but that’s not a given. I’ve blown up ARRI ALEXA clips – that only ever existed as HD – up to 4K and the result was excellent. That has a lot to do with what ARRI is doing in their sensor and the general detail of the ALEXA image. Clearly that’s been proven time and time again in the theaters, where files recorded using ALEXAs with the footage in 2K, HD or 2.8K ARRIRAW have been blown up via 4K projection onto the large screen and the image is excellent.
Don’t get me wrong. I’m not saying you shouldn’t post in 4K if you have an easy workflow (see my post next week) to get there. What I am saying is that staying in 4K versus a 4K-HD-4K workflow won’t result in a dramatic difference in image quality, when you compare the two side-by-side at 100% pixel-for-pixel resolution. The samples below come from a variety of sources, including the blogs of John Brawley, Philip Bloom and OffHollywood Productions. In some cases the source images originated from pre-production cameras, so there may be image anomalies not found in actual shipping models of these cameras. Grades applied are mine.
View some of the examples below. Click on any of these images for the slide show. From there you can access the full size version of any of these comparisons.
©2016 Oliver Peters