On Friday YouTube announced support for “4K video,” meaning video files with a dimensional size up to 4096 x 2304 pixels. Not only does this feature have no practical application in online video, the announcement itself only serves to further confuse an already misinformed public of professional and amateur video publishers.
YouTube purports to make videos “…available in the highest quality possible, as creators intend,” and this is the problem. Bigger is not necessarily better, or more accurately, there is a practical limitation to the gains in perceived quality from video files of a larger dimensional size (notice that I don’t use the term “resolution” here, as the term refers to pixel density, not the number of horizontal and vertical pixels). In fact, going too big will likely introduce a number of performance and quality issues for the vast majority of users.
YouTube mentions that watching videos in 4k requires an “ultra-fast high-speed broadband connection,” but this is actually the least important requirement. While users on slower broadband connections can always wait for enough of the video to download and buffer before watching it, they can’t increase their RAM, GPU performance and display resolution, and those are the source of the real issues.
Most of us these days use digital displays (LCDs, etc.) with native, fixed pixel dimensions. For instance right now I’m on an Apple Cinema display with a fixed pixel dimension of 1920 x 1200. This means that my MacBook Pro’s graphics processor needs to scale any incoming format to the native dimension of my display. The bigger the difference between the incoming format and my display, the more work the GPU needs to do to scale the image. In terms of YouTube’s 4k videos, the end result is that my 1 year old top-of-the-line MacBook Pro can’t smoothly handle the processing, resulting in dropped frames (jerky video playback) and a less than optimal video watching experience. If I switch the stream to 720p or 1080p it looks great, with no perceived difference in image clarity. Actually, I see more compression and motion artifacts in the 4k stream from what I can only guess is non-optimal interpolation from the GPU. You can see for yourself here, but you need to switch to the “Original” format to see the 4k version.
The other issue with YouTube’s announcement is that it makes no mention of bitrate or compression, which is far more important than pixel dimension when it comes to quality. As a video destination catering mostly to non-professionals, YouTube has always used high compression/low bitrates — and their 4k video is no exception. This is why compression artifacts are so noticeable even when watching the 4k version.
Here is a non-magnified section of a screenshot taken of a 4k video on YouTube when played in full screen mode (click the image to see the full screenshot). You can clearly see lots pixelation and compression artifacts, and this is a section with very little motion.
What’s the point of offering higher definition formats if bitrate settings are so low as to introduce significant compression artifacts? This by the way is also a problem with most cable operators: their 1080i and 720p HD signals look terrible due to over compression.
Offering higher dimension video files is technically trivial. It’s simply a matter of adding another transcode profile. But by offering the option to publish ridiculously large video files, YouTube is confusing the public as to what constitutes good video quality and encouraging publishers to upload video files in a format that results in a poor experience for most users.