Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
Blogcast
MillimeterNAB 2005 — Digital Intermediate: Trends and Workflow
Mar 1, 2005 12:00 PM, By Michael Goldman
 Efilm colorist Natasha Leonette employs the Hollywood-based company's proprietary color correction software. |
In a survey of a few technology officials at major companies around Hollywood, Millimeter learned what key trends those facilities are tracking, and what workflow models they are pursuing as DI technology options rapidly evolve.
Among the crucial issues they identified? The pursuit of resolution-independent color correction tools and innovative storage solutions, both of which are key to working and archiving routinely at 4K resolutions, and the need to aggressively advance proprietary color science in-house.
Those surveyed include Hart, Mike Doggett, president of postproduction at Technicolor Creative Services, Burbank, Bob Eicholz, VP of corporate development at Efilm, Hollywood, and Phil Feiner, president of Pacific Title and Art Studio.
Millimeter: What is the biggest change you have made, or will be making shortly, to your basic DI pipeline and approach?
Doggett: Probably the biggest issue or trend is the transition to software-based color correction tools in order to truly be resolution-independent. That's an evolving issue everyone in this industry is faced with. You will need to be able to go back and forth from 2K to 4K. So, we have already added three [Discreet] Lustre systems, and we are examining all the systems out there.
This will mean our colorists, who traditionally have worked on video-based color correction systems, will have to transition into a process where they are basically dealing with files and rendering things. These are steps and processes that would be familiar to a compositor but are relatively new to a colorist. It's a bit of a different mindset, but our guys are already transitioning over and learning those platforms.
Hart: We are definitely looking for a new color correction system that would be resolution-independent, something that would still provide the same features and benefits that the da Vinci 2K [hardware-based system] provides today for our customers. We are looking at every color correction system known to man, but for right now, we'll stay with the [da Vinci] 2K color corrector. I expect that some time in 2005, we will see improvements in the [software-based systems] that will allow us to make a change, but right now, we are in the evaluation stage.
Feiner: [Pacific Title] has been doing DI work on trailers for a few years, and we were a beta test site for the original Colorfront software, but we decided to wait while that technology was picked up by Discreet and turned into Lustre. We wanted to have resolution-independent abilities before we started our DI work, and we now have it. We built our pipeline around Lustre, running on Linux and [Windows NT/XP].
We also wanted to support dual stream HD SDI into a 2K projector, and we therefore waited for Lustre to support dual-stream HD/SDI. Now that has been done, we are up and running with a resolution-independent pipeline, and we can see those images in full color depth at 10 bits on a Christie dark chip 2K projector. We believe this is the best way to approach it, and [at press time], we are in the process of opening a second Lustre-based DI room.
Eicholz: Our core color correction software [a proprietary version of the original Colorfront system] is already resolution-independent, so we do not anticipate having to make a major change to accommodate 4K work. In fact, we have already done the Spider-Man 2 DI [see the August, 2004, Millimeter], which was 4K except for the visual effects, and more recently, the Ocean's 12 DI, which was done on a full 4K data path, including visual effects.
A bigger change for us has therefore been adding more storage. In the last year, I added 128TB of storage to our facility — Ocean's 12 alone required between 30 and 40TB. So, we are now ready for a full 4K pipeline, which I believe will be the standard for the industry.
 Pacific Title and Art's DI color correction suite uses its Discreet Lustre to output dual-stream HD SDI into a 2K projector. |
Doggett: The reason to work at 4K is for archival purposes, not for viewing purposes, and that is how we view it — view proxies at 1K or 2K, and then render scans at full 4K. As time goes by and processors get faster, realtime 4K will be an option, though I have no idea when that might happen. But for now, the model is a hybrid model — scan at 4K, work using resolution-independent software, view it at a lower resolution, and then render and archive at 4K.
Eicholz: At 4K, data management is the same, but the files are bigger, and rendering takes longer. Therefore, we had to beef up our processing PCs so that we do not have to compress files. 4K will eventually be the industry standard, but right now, it is simply the best way to archive — the higher resolution you can use for archiving, the better. As long as your facility is not based on hardware confined to 2K DIs, and your network is scalable, you simply have to incrementally add more storage, and that is what we are doing. I expect to add another 65TB in the coming months.
I expect that within about two years, the majority of A-list feature DIs will be done at 4K. But there will also be a second tier of independent films and lower-cost features that will remain at 2K.
Hart: There is this notion that more is better, and everyone wants to be as close as possible to the theoretical resolution of film. But, frankly, if you are not working in 70mm, I don't see where there is a huge difference [in terms of viewing].
In fact, the compression tests I have seen using the JPEG 2000 [image coding system using new compression techniques based on wavelet technology] are quite impressive. They showed the differential of information available in the signal between a 2K and a 4K image, and frankly, the difference is so miniscule at that point that I kind of think we should just get out there and make movies. People want to do 4K, but the truth is it's not yet practical to do it that way, as not all the toolsets are viable for 4K right now. Modern VideoFilm is a leading-edge company, and when it is possible in a practical sense, we will offer 4K in every facet of our workflow.
Feiner: In terms of viewing, there is a difference between 2K and 4K, but it is content-driven, depending on the project. If you use fine-grain film to acquire an image with a locked-off camera with lots of close-ups, details, and landscapes, you can see slight differences in sharpness between 2K and 4K. If you are panning and using a quick-motion camera, you won't notice. The migration will take some time, but the point is, for a facility like ours, it makes sense to be resolution-independent now so that you can work at 2K, 4K, 6K, whatever is needed.
 The color timing cinema suite at Modern VideoFilm in Burbank. |
Hart: We use Xyratex and JMR systems, installed by Bright Systems [Nevada], and they work quite well for our current needs. But I have a lot of concern about storage as we expand. I believe we have something like 86TB of spinning disks right now, and that doesn't include the Dylan storage on our [Quantel] iQ systems, which probably takes us well over 100TB.
The more spindles you have spinning, the more problems you can run into. I am personally examining ways to push disk performance issues into the background by finding some kind of device that would have smart memory cache so that when you get a slow disk, you don't worry — you can still provide data throughput to your machines without being dependent on the disks, and being able to rebuild the discs on your schedule, rather than in some emergency situation. You find yourself rebuilding RAIDs quite a bit in this business, no matter whose disks you use.
Doggett: We use a hybrid approach with combinations of SAN-based storage in concert with platform-specific local storage, so having it all compatible so that we can use the best combination of tools is always a concern. But I find the advent of removable storage very interesting, and we are looking at several different options for increasing our ability to move drives around efficiently from point A to point B.
Eicholz: We generally use SGI ATA storage for our SAN, which is slightly slower, but a third of the cost of high-speed storage. It's been adequate up until now. I've heard of a new drive for serial ATA storage coming out that is denser, with about 400MB in each drive. That will decrease the amount of space our drives take up.
But other than that, it's just a real-estate issue. We have a large SGI CFX network, and the beauty of that SAN structure is that we can keep adding to it. The other beauty of the system is that it acts like one big disk drive — one big server. Anyone in any DI room can access image data anywhere. I think this scalable model is the best way to do big 4K shows.
Millimeter: What about the issue of color science, LUTs, and industry standards? Do you see any need for standards as DIs become more ubiquitous, or are proprietary LUTs part of the unique product that facilities use to set themselves apart?
Eicholz: It's still very early in the life cycle of this [DI] industry. Companies like ours can't wait for a full set of standards to be put into place in order to get the color right. Therefore, we all work on our own color systems.
That said, we do support the idea of standards and are actively participating in their development. Ideally, systems incorporating industry standards will be developed, but with provisions for incorporating each facility's “special sauce.” Filmmakers and studios want standardization. But they also want the latest developments to give their films the best possible look. For now, a blend of industry standards and proprietary software is the way to go.
Doggett: It's a creative process, so you can't simply standardize everything. In the simplest terms, when you are color-grading something, at the end of the day, you want the digital side of the screen to look the same as the film side of the screen. But it's all a creative choice for the filmmakers. So, simply, if off-shelf tools are not available to fit the needs of all those different scenarios, we will build our own LUTs as we have always done.
Feiner: [Kodak] Vision film processed and printed through Deluxe looks different than processing and printing through Technicolor. All major labs create their own look on purpose [for film stocks], and the filmmakers decide what they want to use and how they want it to look. We do not want to dictate the look to the cinematographer. We want to give them whatever they desire.
So we need LUTs that give them that kind of a choice, and other proprietary tools, as well, for that matter. We have our own, automated dust-bust system, for instance, in our pipeline, and we also wrote our own database for managing the DI process in-house.
Hart: I don't see how it would be possible to have industry-standard lookup tables when we are in a world with so much variability. Competition is also a significant factor.
We chose the Truelight [color management] system from FilmLight, so we have the ability to build our own 3D LUTs as needed to match various [viewing] devices or to provide color space conversions for different delivery requirements.
One thing that does concern me as companies like ours create new LUTs for our purposes, however, is how to maintain control of that intellectual property. If it leaves your possession and goes into someone else's machine, then you spent a lot of money to help some other individual or company. We prefer to install our LUTs in Truelight hardware to keep control over our intellectual property.
Continue the discussion on “Crosstalk” the Millimeter Forum.