Another Red Day

Yesterday, I attended the Red Day presentation done at Ren-Mar Studios (oops…. I meant Red Studios Hollywood – old habits die hard in this town..). Saw a lot of familiar faces, and met a lot of new ones. I especially enjoyed seeing Assimilate’s use of dual Red Rocket cards to provide live playout of stereoscopic Red material with full debayering in real time. That alone has a lot of potential that I hope to explore.

This morning, a discussion on the CML centered around things Red (and possibly other companies) might be able to do that would be truly revolutionary and useful. It seems to many that most of the talk involving new digital cameras centers around things like improved dynamic range and resolution – important things to be sure, but ones that are really incremental improvements, not revolutionary changes. To be revolutionary, something has to be presented that accomplishes something that cannot currently be accomplished, or at least accomplishes it in a new way that changes the way one looks at the problem. It is very helpful if that change is also useful, in terms of either making a task more efficient, or eliminating costs associated with doing things using the current methods. I’ve got some things to suggest that I think might be revolutionary and useful. All relate to characteristics of the current Red systems that are often criticized, such as its use of a proprietary file and compression format, the need to supply personnel and systems for backing up files at the time of production, the complications involved in maintaining a consistent color path for dailies, the need to constantly transcode camera files, and the lack of a proper archival element. Here are some of them:

1. On-camera redundant recording. The modular nature of Red’s new designs allows for a lot of innovation. I would suggest that one of the first could be a double-redundant recording module, with, say, 2 CF card slots. They would record directly from the camera head, the same way the existing unit does now. But it would immediately provide 2 additional “original” copies and thus eliminate the need for multiple backups on the set (and the systems and personnel needed now to do that). To minimize media costs, permanent backups and on-line working files would then be done either by a post facility, the production office, or editorial, where personnel and equipment is already in place and thus doesn’t impact production, either in terms of time or cost. Once that’s done, the “shuttle” cards would be wiped and returned to set. Ideally, it would be a smart module, with its own processing, that could also trigger CRC checks as soon as you hit Stop, and report any problems to you on either a built in screen, a wireless connection, or their new “Redmote.”

2. A module for simultaneous creation of files for editorial. Say, a module that takes a video feed from the SDI output on the already proposed output module and records, in real time, using either DNxHD/MXF or ProRes/Quicktime, depending on what editorial is using. It could record on, say, an SD card (or maybe two for redundancy, as mentioned above), using an internal solid state recording module as a buffer if necessary, and completely eliminate the need for file conversions at the dailies stage. It would by definition have any color settings you’ve used for monitoring already baked in, it could include sound if you’ve got it, and it would get all of its metadata via the modular bus.

3. An archival format based on open standards. This could consist of a utility supplied by Red that would decompress the existing R3d file back to its RAW components and, without debayering, either recompress using an open standard, perhaps JPEG2000, or write out an uncompressed RAW image, including only the selected frames in either case. They could then be saved as essentially digital RAW separations (that’s the form they would be in anyway if there is no debayer), with some kind of sensor definition tag included in metadata for proper reconstruction at any time in the future. If done right, it could be argued and probably proven that such a process, if controlled by the manufacturer, would provide a very accurate record of the original data in a form that is currently unavailable (and could, in fact, ultimately benefit from more advanced reconstruction and processing techniques), and that could and would serve as a very compact, reliable, and accessible archival element.

Item 3 would clearly have to be done by Red directly. The other two could be done by third parties if Red would see fit to license or release hardware design criteria in the same way they provided the SDK for software development, or if they would contract the work out. I prefer the licensing approach because it would invite competition, which always gets better work done, and usually more quickly.

Of course, the most revolutionary and potentially helpful change would be for Red to stop the unending and constant revision of their software and firmware. The unending updates, the speed with which they’re released, and the fact that many of them tend to break what came just weeks before, leads many to the conclusion that their products are not suitable for longform projects which require maintaining some consistency over a longer period of time – something you get with film or SR videotape recording, mediums that don’t change every other week. Many of their users seem to be perfectly willing to spend hours and hours on RedUser and change firmware moments after a new version shows up, no matter how many times Red states that it is “early beta” quality. Professionals require more long term stability and predictability in their tools, especially when production will take place over months and not days. This, to me, still remains one of the primary objections to the use of Red by network television programs and many studios, along with the items mentioned above. I understand the excitement with which they approach the ability to change and fix things, but at some point, they need to establish some stability. Perhaps the ultimate release of the new products will lead to this. I hope so.

And, by the way, if anyone from Red is reading this (I hope someone is..): Please don’t take offense at my use of the Ren-Mar name. Not only did I work there for a year (during the first season of “Ally McBeal,” on which I was visual effects supervisor), but I still refer to the Sony Studios lot as MGM. But then again, so do a lot of other people in the industry..