Because the 35mm footage would serve as the benchmark for the tests, Sowa began by color-correcting that footage. He sat with the cinematographer from each test scenario and did two passes on the film footage. First, they did a “best-light” color-correction with no secondaries, windows, dynamics or keyframing; because the cameras often moved through changing lighting conditions during the scenes, Sowa optimized the timing for one moment. Then, they did a final grade using the full DI toolset except for noise-reduction and sharpening tools. (The CAS guidelines forbade noise reduction and sharpening throughout the entire workflow, both on set and in post.)
With the look set for the film-originated footage, Sowa set to work on the images from the digital cameras. Again, there were two passes: the best light (optimizing for the same moment chosen for the film camera) and then the final grade, which focused on making the footage match the look of the film footage as closely as possible, regardless of how many adjustments were needed. At no time were any of the digital cameras matched to one another or even shown side-by-side. “It was the goal of the PGA and the ASC that no electronic camera would be directly compared to any other electronic camera,” says Bill Bennett, ASC, who was the on-set cinematographer for the Arri D-21 and sat in on many of the grading sessions.
The camera manufacturers were allowed into the DI suite, but only while the footage from their particular camera was onscreen. “We decided to do the color-correction on a scene-by-scene basis,” says Bennett. “Most of the time, the manufacturers were able to send a representative for each timing session. They had to come back day after day and wait in the lobby until their camera came up.” He adds that he found it valuable to see the whole post process. “Cinematographers usually show up for the color-timing session, but we don’t often see the data-transformation aspect of it, or all those other peripheral things. To learn about how that happens was very helpful.”
The final film print and DCP were first presented at a pair of events in June, the PGA’s Produced By Conference and a private ASC event. One of Clark’s initial impressions was that “we now have digital motion-picture cameras that are refined and capable of producing some extraordinary results in today’s DI workflow environment,” he says. “You can use a top-performing digital camera without necessarily being handicapped by choosing that camera over film. A few of these cameras are demonstrating that they are able to adapt to existing, film-centric DI workflows and do so very effectively.”
However, says Stump, no one should assume that enough time and effort can make images captured by any digital camera look like film. “That’s not the right conclusion to draw from this,” he says. “We’re almost there, but we have to keep making more demands of the manufacturers to refill our toolbox with electronic tools. There are still many things you can do with film that you can’t do with electronic tools. The manufacturers have been listening — that’s why all these cameras look so good — but let’s not let anyone off the hook. This test points out where we can improve the entire imaging chain, from acquisition to display. It isn’t good enough yet.”
At the Produced By Conference, McCreary mentioned a few of the improvements manufacturers still need to make. “We would love optical viewfinders in the digital cameras — in our assessment, only the Arri D-21 and the Arri 435 film camera had optical viewfinders,” she said. “We also asked for true 2K and 4K cameras in terms of both sensors and storage — no compression. Another request was that the manufacturers help us define and implement metadata standardization.”
Bennett agrees that manufacturers need to provide an easier way to input the metadata on set and tie it to the file itself, ensuring that it is always present and retrievable in post. He notes that the CAS incorporated a primitive sort of metadata in the form of color Post-It notes that were always visible in frame; a different-color Post-It was used for each camera. “That way, we knew we’d always be able to tell which camera it was,” he says.
Stump, who also chairs the ASC Technology Committee’s Metadata Subcommittee, believes metadata is one of the last great places to save money in filmmaking. “Producers and the studios expend an enormous amount of effort to squeeze every nickel out of the production budget and get it on the screen,” he says. “If they only knew how much is being squandered by inefficient workflows that could be automated by metadata! People just don’t recognize it for what it is. It’s going to take the whole community to implement a rich, automated, uninterrupted stream of metadata, but if the entire industry pursues it, the production community will realize big savings they never knew were there.”
Stump also notes that although 4K finishes are uncommon, there is good reason to push camera manufacturers in that direction. “It was announced in June that Texas Instruments and all the projector companies are going to be supplying 4K projectors, which means 4K exhibition will eventually be ubiquitous. That will give us 4K projection, 4K DCI standards, and a fairly nice movement toward 4K finishing, workflows and color correctors. The only thing we don’t have is a true 4K digital-acquisition device that supplies co-sited RGB pixels at 4K each. Everyone has to realize that ‘good enough’ isn’t good enough.”
Based on the CAS, though, there is a great deal of optimism about the manufacturers’ level of engagement and commitment to the industry. “It’s quite a tribute to the manufacturers that they all stepped up and participated fully in the CAS,” says Bennett. “In the post phase, we got the distinct impression that they were learning as much or more than we were about blending their cameras into established workflows. They all made some tremendous realizations.”
Already, some of the camera manufacturers are beginning to offer their own LUTs to bridge the gap between the images their cameras capture and the film-centric DI workflow used for the CAS. “The more camera manufacturers have to stew in that juice, the better they can appreciate why digital is not ubiquitous as an acquisition medium yet,” says Stump. “In that respect, the CAS was a huge success.”
Bennett emphasizes that the CAS is only a starting point, and before commencing any project, filmmakers should “test as much as possible, and carry those tests all the way through to the way in which it will be distributed — film print, television, Blu-ray DVD. You must test all the way through the process to discover the limits of each imaging system and then work within those limits. All imaging systems have limits. Even with paint on canvas, artists had to learn what they could and couldn’t do. Then, applying their skills, they could make beautiful images.
“As digital acquisition evolves, we’re learning what these cameras’ strengths are, and the cameras are being used for those strengths,” continues Bennett. “That’s the biggest benefit the CAS can offer filmmakers: we can use the cameras in the situations to which they’re best suited.”