While digital motion picture cameras have been around for a while, the number of productions making serious use of them has been limited. Notable exceptions include "Cold Mountain," but the exception looks like it could soon become the rule. Several new offerings promise to overcome the workflow issues that have arisen. Improved resolutions, single-chip architectures, and better shoot-to-post data transfer may very well vault the technology into the mainstream.
With any change come new challenges. In the same way that parents move from diapers to dating and tantrums to tuition, a move to all-digital production will shift the required facility focus to the storage and networking domains. Gone will be the days spent fighting with the lab over time and costs billed. Replacing them will be IT-related issues of managing digital footage as it arrives from the set.
So, let's start at the beginning from a post perspective. Digital film recorders employ one of two transports - video or data. This is to say that they either record film frames via a dual-link HD-SDI connection, or using a specialty IT-centric connection that accepts pure data. Whatever storage system a post facility chooses to implement will have to accommodate the format in which the data arrives.
All major cameras on the table today, save one, are employing the video method of getting the data off the camera. Some choose to simply output at HD resolution, but there?s a growing resistance to such a limited image size and bit depth. Others are looking to encode larger resolutions into the dual link HD-SDI stream similar in fashion to the film scanners that support the high-speed data link (HSDL) protocol. A quick look at the transport on the three latest offerings in the digital film domain will be useful. The Arri D20 and Panavision Genesis use video-based transports (dual link HD-SDI), whereas Dalsa with their Origin camera takes an entirely new approach in how it outputs the data. Instead of using a video-based transport, it employs an Infiniband-based connection to its recorder.
For those not familiar with Infiniband, it is an extremely high-speed network interconnect that can accommodate both control information and data. Currently shipping hardware is demonstrating in excess of 500MB/sec. of sustained performance, or more than double the rate that can be carried as data over a dual link HD-SDI connection. All this translates into an ability to move higher resolution images at higher frame rates - both into the recorder and back out at the post facility. The Origin can output 24fps 4Kx2K in realtime, something which will not be possible using a video transport.
Once the film has been shot, the requirement now exists to get the data directly back to the post facility. No lab or scanning facility is in the loop anymore. This is huge on many levels. The dailies process will be redefined as on-set viewing and simplicity of transfer will likely eliminate the need for selects altogether. A project can move from production to post a quantum leap quicker.
There are two reasonable methods by which the material can make its way back. The first is removable media and the second is high-speed telecommunications.
Removable media will be in the form of videotapes or hard drives.
Videotapes will clearly be the simplest and most cost effective method, but would only be appropriate today when the recording device is a Sony HDCAM SR, which supports dual link HD-SDI. With that said, given the near 2:1 compression that the data undergoes for this format, many may be reluctant to make use of it.
Hard drives can either be the drive set the recorder uses, or some external drive set (e.g. Firewire). The native recorder hard drive set will most likely be the transfer method of choice due to the speeds at which data can be extracted without the overhead of first dumping to an external drive set. This is again where the decision that Dalsa made to employ a high-speed networking interface will likely pay off as their recorder can be emptied much quicker into the post facility vault, getting the facility working sooner and the drive set back into the field.
From a telecommunications perspective, there's no technical reason that the recording devices (with the exception of videotape-based methods) couldn?t be connected via a short or long haul network to transport the data electronically directly to the post facility. However, the current cost of such a connection, given the time constraints and the size of the data set to be moved, will have to come down before it can become a cost effective approach.
Once back at the post facility, the material needs to be extracted, catalogued, and then likely written to yet another backup device for security's sake. One very interesting point to note about both the Arri and Dalsa cameras is that they can record in what?s called the "Bayer" format which is akin to a digital negative in that it's the pure data as seen by their respective silicon-based sensors. The bit rate for Bayer data is only 1/3 that of traditional film as stored in 10-bit log format, so it is in a sense a compressed way to carry film frames around. Of course these film frames will have to be decompressed (or processed to use a film term) on demand to be useful in a post context, but given the load that will be taken off the facility network, it will certainly prove to be a very efficient way to work.
In summary, given the level of celluloid-related headaches that exists today, there's no question in my mind that digital film shoots are positioned to become real, and that post houses need to get going now in augmenting their infrastructure such that they can accommodate the volume of material that will be arriving directly from the field.