Analogue Wayback, Ep. 20

D2 – recursive editing

Video production and post transitioned from analog to digital starting in the late 1980s. Sony introduced the component digital D1 videotape recorder, but that was too expensive for most post facilities. These were also harder to integrate into existing composite analog facilities. In 1988 Ampex and Sony introduced the D2 format – an uncompressed, composite digital VTR with built-in A/D and D/A conversion.

D2 had a successful commercial run of about 10 years. Along the way it competed for marketshare with Panasonic’s D3 (composite) and D5 (component) digital formats. D2 was eventually usurped by Sony’s own mildly compressed Digital Betacam format. That format coincided with the widespread availability of serial digital routing, switching, and so on, successfully moving the industry into a digital production and post environment.

During D2’s heyday, these decks provided the ideal replacement for older 1″ VTRs, because they could be connected to existing analog routers, switchers, and patch bays. True digital editing and transfer was possible if you connected the decks using composite digital hardware and cabling (with large parallel connections, akin to old printer cables). Because of this bulk, there weren’t too many composite digital edit suites. Instead, digital i/o was reserved for direct VTR to VTR copies – i.e. a true digital clone. Some post houses touted their “digital” edit suites, but in reality their D2 VTRs were connected to the existing analog infrastructure, such as the popular Grass Valley Group 200 and 300 video switchers.

One unique feature of the D2 VTRs was “read before write”, also called “preread”. This was later adopted in the Digital Betacam decks, too. Preread enabled the deck to play a signal and immediately record that same signal back onto the same tape. If you passed the signal through a video switcher, you could add more elements, such as titles. There was no visual latency in using preread. While you did incur some image degradation by going through D/A and A/D conversions along the way, the generation loss was minor compared with 1″ technology. If you stayed within a reasonable number of generations, then there was no visible signal loss of any consequence.

Up until D2, performing a simple transition like a dissolve required three VTRs – the A and B playback sources, plus the recorder. If the two clips were on the same source tape, then one of the two clips had to be copied (i.e dubbed) onto a second tape to enable the transition. If you knew that a lot of these transitions were likely, an editor might take the time before any session to immediately copy the camera tape, creating a “B-roll dub” before ever starting. One hourlong camera tape would take an hour to copy. Longer, if the camera originals were longer.

With D2 and preread, the B-roll dub process could be circumvented, thus shaving unproductive time off of the session. Plus, only two VTRs were required to make the same edit – a player and a recorder. The editor would record the A clip long in order to have a “handle” for the length of the dissolve. Then switch on preread and preview the edit. If the preview looked good, then record the dissolve to the incoming B clip, which was playing from the same camera tape. This was all recorded onto the same master videotape.

Beyond this basic edit solution, D2’s preread ushered in what I would call recursive editing techniques. It has a lot of similarities with sound-on-sound audio recording innovated by the legendary Les Paul. For example, television show deliverables often require the master plus a “textless” master (no credits or titles). With D2, the editor could assemble the clean, textless master of the show. Next make a digital clone of that tape. Then go back on one of the two and use the preread function to add titles over the existing video. Another example would be simple graphic composites, like floating video boxes over a background image or a simple quad split. Simply build up all layers with preread, one at a time, in successive edit passes recorded onto the same tape. 

The downside was that if you made a mistake, you had to start over again. There was no undo. However, by this time linear edit controllers were pretty sophisticated and often featured complex integrations with video switchers and digital effects devices. This was especially true in an online bay made up of all Sony hardware. If you did make a mistake, you could simply start over using the edit controller’s auto-assembly function to automatically re-edit the events up to the point of the mistake. Not as good as modern software’s undo feature, but usually quite painless.

D2 held an important place in video post. Not only as the mainstream beginning of digital editing, but also for the creative options it inspired in editors.

©2022 Oliver Peters