The Zombie Stack Exchanges That Just Won't Die
Is there still an economic argument for preserving born-digital video at less than 4:4:4?
Jerome McDonough 2004 argues for 4:4:4 (chroma subsampling) or nothing, although he points out that this is sometimes dangerously expensive.
Since disk storage costs have gone down (and up) since then, what's the rationale at this point, for keeping anything less than a "pristine" set of workfiles?
For clarity, I'm referring to what George Blood 2011 describes as "Digital Source, media independent 'file based'", and not things that are recovered from tape or such.
Al Matthews
I may be wrong, but aren't modern lossless frame compression algorithms (such as JPEG2000) providing better compression ratios than chroma subsampling, without any loss of quality?
Sure you can compress chroma plane with less bitrate, that will still preserve a lot more data than "dumb" subsampling, and possibly boil down to smaller compressed size than subsampling, too.
The compression speed might be an issue, of course.
There is no economic argument for preserving something only because it exists. It depends fully on what is in that video.
I don't understand what you understand by Pristine (AFAIK it's the main city of Kosovo province), but generally, the videos of high value should be stored in the highest quality available. Example of such videos are documentaries, reportages etc. When it comes to lower quality videos, such as thousands of hours of video from the foreign trip, if they were too low quality or too redundant to use for documentary/reportage, storing them in reduced quality is better alternative than deleting them.
For such materials as private wedding videos, when the people interested in them no longer live, I think it would be enough to use heavy compression and resampling and archiving them with family and temporary tags. The may be of value for future anthropologiests or historicans.
As general rule, I would never trust someone who is going to describe the whole complex process with one single sentence. Video has no value because it is a video but because of its content. So the economic criteria can be applied only when that value is taken into account.
I am assuming that the George Blood paper you are referencing is Refining Conversion Contract Specifications (2011-10-01). The section that covers "Digital Source (media independent "file based")" video is dealing with what is also referred to as "born-digital" video. These video files are born with specific characteristics, chroma subsampling being one of them. These characteristics should be considered significant properties of those files, for preservation purposes. Chroma subsampling refers to color compression, just as data rate refers to temporal compression. When we talk about "uncompressed" video, we are usually talking about temporal compression. When an archive chooses "uncompressed" as a target video spec for reformatting analog material, they often still have color compression, typically 4:2:2. We accept this degree of color sampling because the human eye is less sensitive to color than it is to luminance.
But since you referenced born-digital, the issue is slightly different. For the vast majority of these files, the native chroma subsampling is 4:2:2, 4:1:1, or 4:2:0. There are only a handful of cameras on the market today that are capable of producing 4:4:4 video, and those are generally high-end, professional digital cinema cameras.
So, in short answer to your question, if we only accept 4:4:4 files we will not be preserving 99%+ of the video content being produced today. And that is unacceptable.
Digital video production and preservation has changed dramatically in the past 9 years since Jerry MacDonough wrote about 4:4:4 video. At that time, digital (file based) preservation of video was very theoretical, and certainly not practical. That's changed to the point where today it is the ONLY option for video preservation. And costs have dropped to the point where it is feasible and realistic to create temporally uncompressed video. But generally speaking, there is no advantage to 4:4:4, and few systems that support its creation at this stage. That may change in the future, but for now, we are just not there.