title: The disintegration of a linear timeline.

We’ve lost something important in the way that we capture and archive video.

I was going through the somewhat tedious process of digitising and archiving some old VHS tapes of mine recently. Video that was mostly captured on cheap camcorders and has been carried from house to house over the past 25 years or so in a cardboard box.

A few things stood out to me from this process.

Firstly, how well these old tapes have held up over nearly three decades. I haven’t been particularly diligent or precious about how they’ve been stored - they’ve lived everywhere from garages to closets to storage units through every Australian season, probably surviving temperatures close to zero degrees Celcius, and as high as fifty or sixty degrees celcius. Despite these harsh conditions, they’ve stood the test of time well. Of the dozen or so tapes I had, I was able to successfully digitise all of them.

It’s probably the greatest volume of electronic information I’ve managed to retain from that period of time, when hard drives were typically measured in single-digit gigabytes, if you were fortunate. Compressed, it probably represents several hundred gigabytes of information that has survived.

The second thing I noticed is how watchable it was. We take 4k, HDR video with stereo sound and Dolby Atmos on our phones for granted today - by comparison, this video would be lucky to have 200 lines of usable horizontal resolution and a single monophonic audio track with 10khz of bandwidth. And yet, despite these limitations, it didn’t detract from my enjoyment of revisiting this old footage.

In fact, the look and feel of the video was a critical part of my memory of it, and the way I experienced it at the time. Had it magically sat in a box and improved in quality, it would lose its authenticity as a memento of that point in time - the questionable quality is part of the memory.

This is something I’ve been reflecting on lately when considering the relative merits of RAW vs JPEG for still images. I’d always defaulted to RAW as soon as I had cameras capable of supporting the format. RAW should be objectively better as it captures more information and allows more latitude for future processing. But the look of a JPEG captured out of camera, with processing and colour science already applied, somehow comes to form the definitive version of how I later feel that photo should look. I found this was the case when reviewing some old travel photos from an iPhone, only to find that Apple Photos wasn’t correctly applying the metadata. It was immediately obvious that the image didn’t exist as I remembered it, and it felt foreign and strange. I feel like the same would be the case for these same old videos if I attempted to change their aesthetic.

But the third thing I noticed, which is what prompted this article, was how we seem to have lost the concept of linear video.

Virtually all video recording media once linear by nature. The first video camera I used, which I found in my primary school’s sports locker, probably dated from the 1960s and used linear reel-to-reel tape. Video was recorded to it in a strictly linear fashion: there was no editing or deleting, only fast forward, rewind, play and record. Editing of video was a slow, time consuming, expensive and cumbersome process, which either required multiple recorders, or physically splicing tape.

Linear tape was superseded by VHS and Beta, which used a helical scanning system to effectively increase linear tape speed, whilst keeping transport speeds low and improving efficiency by recording diagnonally across the medium. These formats were joined by Hi8 and VHS-C for portable applications, still operating in the analogue domain. Even the move to the first generations of digital formats still recorded linearly, with clever implementations that allowed segments to be overwritten without breaking the digital stream.

All of these formats meant that multiple segments of video were viewed linearly as they were recorded on the tape. Cleanly removing or editing mistakes was time consuming, if possible at all. What was captured to tape, or later, optical disc, was not only a capsule of the events themselves, but the way they were captured - the process of recording and manipulating the tape itself became part of the record. Watching back those old tapes, the breaks and mistakes and portions where one scene was careless overwritten by another were as much part of the memory as the scene being memorliased itself was.

But the biggest piece we’ve lost from that era is the act of watching a series of linear recordings sequentially. As me moved the linear recording to file-based recording, used today by practically every digital camera, every scene has become its own file, often interspersed with photos. Most video playback software will play a single file at a time; those that do will often playback consecutive files in non-standard ways. We relive these memories one ten-second snippet at a time, jumping through media chaotically and without a strong sense of definitive order. Two different people, watching the same collection of clips, will experience that set of clips completely differently, and with different context, depending on the method they use to play those clips.

It also probably means that a large amount of capture video goes unwatched. Almost every time I’ve observed someone replaying video from their photo library on their phone, they hunt and peck through some random clip in time, based on its thumbnail, watch it, and return to the gallery to find another clip to view, which often isn’t the next consecutive clip.

There are of course solutions to this: individual video clips can easily be concatenated end remuxed to form a single definitive stream that forces a viewing order. There are two problems with this approach though, that I can think of. One is that it takes extra time and software. Most operating systems and devices (phones/tablets) require creating a project, importing the clips, and laying them in a timeline in the desired order, before finally exporting a new stream. Not complex, but more work than simply leaving them in an album on one’s phone.

The second issue with this is the loss of metadata. Each clip has important information such as the date and time it was recorded, as well as proprietary camera information. When an edited clip is produced, that metadata (and context) are lost. It is then incumbent upon the editor preparing the video to preserve that data (and context) through other means, such as titles or lower thirds.

This is where I’ve developed a renewed interest in formats such as AVCHD, which I’d previously dismissed as being technically inferior to other transport streams such as MP4 and MOV. It seemed to be more complex - and more proprietary - with seemingly artificial limitations such as on bitrate that other formats didn’t have.

But it was specifically designed to solve the problems I described above.

AVCHD isn’t that disimilar in many ways to the MP4 and MOV contained formats that have become ubiquitous: it still uses the h.264 standard for encoding video. But internally, it uses the MTS (MPEG Transport Stream) container format, which is commonly used in professional video authoring, Blu-ray video and terrestrial video broadcast. Where MP4 and MOV are designed to be flexible and adaptable to vendor requirements, MTS is tightly specified.

AVCHD also captures data about the temporal arrangement of streams through a system of playlists. This allows individual clips to be combined into a seamless linear stream for playback, whilst still allowing random access to individual clips and