As promised-slash-threatened, here’s a not-brief write-up of some of the technical aspects of making The Nines.
The movie is made up of three distinct sections, each of which was shot in a different format. That was always part of the plan. The movie is really like three short films back-to-back, and each of them needed its own look and grammar.
For Part One, my hope was to shoot HD. Even as I was finishing the script, I’d begun a conversation with Mike Curtis at HD for Indies about the potentials and pitfalls of various cameras and workflows. In case you ever doubt my the extent of my geekery, check out the four pages of flowcharts I made to map out the process we came up with: nines_workflow.pdf.
Mike was a big help in letting me talk through, and think through, my goals and priorities in the technical details of shooting movie. But how much of his advice did we end up using? Almost none.1 As it turned out, we didn’t shoot HD at all.
I really thought we would. But our d.p., Nancy Schreiber, quickly convinced me otherwise. She’s no Luddite — she’s won awards for her digitally-shot features. But when she visited our main location, with its vast expanses of glass, she made it clear that any savings we would have gained from shooting on HD would be lost by the extra time and expense it would take to control the light.
So out went the plan for tethered cameras and hard drives. Instead, we shot two sizes of film, and standard-def video.
How we shot
For Part One, we shot Super-16, a format that’s come roaring back with the introduction of great new stocks by both Kodak and Fuji.2 The cameras are solid, and the workflow is straightforward. There are downsides: negative dirt (the white blotches) gets gigantic when blown up to full size, and it has all of the hassles of film (loading, storing, nightly film drops). But I suspect we’ll continue to see indie features and television shooting on Super-16 for some time. It’s a great image for the price.
For Part Two, we shot standard-def 24p video, using mostly the SDX-900.3 Why not HD? For starters, this section is meant to be a reality show, so we didn’t want it to look too much like film. But the main reason was Nancy’s experience shooting the Lisa Kudrow show The Comeback. With all its detail and resolution, HD can be pretty harsh on actors. That’s why Nancy chose to use standard-def on both The Comeback and our movie.
It was the right choice: sharp but not harsh. It’s a great contrast from the Super-16 in Part One, and still looks solid when projected on the big screen. In fact, I’d push any no-budget indie to keep 24p SD in the mix, particularly if it’s mostly people talking. Resolution is over-valued.
Part Three was shot the most conventionally, using 35mm. However, we shot 3-perf rather than 4-perf.4 It’s a significant cost savings, and since we knew we were never going to cut negative, there was no reason not to use it.
I sort of buried the lead there, so let me put that last sentence in context. We knew we were never going to cut negative, because on a practical level, there was no negative.
Yes, the film had negative, and it was stored in boxes at the lab, but we knew we were never going to touch it again. And the original videotapes from the camera were dutifully labeled, but will never be put in a deck. That was the decision we made quite early in the process: no matter what format we shot on, our “masters” would be the HDCam tapes we got back from the lab.
The Super-16? HDCam. The 3-perf 35mm? HDCam. The SD video? Up-rezzed to HDCam.
Editing
Picking one format made post-production vastly simpler, because it made no difference what something was shot on. The Avid ingested everything off the HDCam tapes, completely oblivious to what form it began its existence.5 loves Avid, and he has an Oscar nomination. So I deferred to him. And you know what? Avid is great, too. Motion picture editing is clearly benefiting from having two strong competitors in the field. Note that HDCam decks are expensive to rent. We couldn’t afford to keep one for the run of the show, so we just got one as needed to load in footage.
Did we make some mistakes? Yes. We built the show as true 24p, when we should have chosen 23.976, which would have made hooking up inexpensive monitors much simpler. But I felt some vindication upon learning that $100 million movies were hitting the same snags we were.
In terms of visual effects, almost everything for the movie was handled on HDCam or D5, with a lot of lower-res QuickTimes emailed back in forth to show work in progress. Producers’ screeners were simple DVDs.
After a few months of editing, when it was finally time to output the movie from the Avid, we rented the expensive D5 deck. From that point on, that tape was like our negative. We would clone from that for color-timing and other work. (Once color-timing was done, that D5 became our new negative.)
I’m skipping the various formats we used for sound, because it was honestly over my head. Suffice to say there was always some way to make this frame rate link to that frame rate, although I always time-budgeted an extra hour whenever we had to deal with something other than pure picture. My advice to filmmakers is to make sure that your post supervisor, sound designer, editor and composer get friendly, and let them sort it out.
The DI
In an earlier article (“Digital filmmaking and the paradox of choice”), I argued that all the great new tools available to filmmakers can cause a kind of paralysis, where a surplus of options means nothing gets decided. I was determined not to fall into that trap, which is one reason why we front-loaded a lot of the color correction in the movie.
Basically, every movie goes through color correction twice. The first time is when each day’s work is processed into dailies. It’s called telecine, and the goal is to get the video footage looking somewhat like the final film, so that you get a sense of it while editing. In conventional movies, the initial telecine isn’t a make-or-break step, because the “real” color correction will happen later, working off the original negative footage. (Either a conventional timing, or a digital intermediate. (DI))
With our film, the dailies were our negative, so we decided to get them very close to the final look of our picture in the initial telecine. That’s actually counter to a lot of opinion about what you “should” do. Many experts will tell you to keep your negative fairly neutral in order to allow yourself the widest latitude down the road. The problem is, you’re just pushing back the decision process, and you pay for it. Literally: telecine is a relatively fixed cost, while the hourly rate of a DI room can kill you.
So Nancy and I picked our looks for the three sections off of one day’s test footage. As you can see from the trailer, one section (Part Three, the forest exteriors) is aggressively desaturated, and that was decided in the initial telecine.If I’d changed my mind a few months later, we could have rescanned the film, but it would have been incredibly costly. By locking down our decisions early on, we could spend our time in the “real” DI tweaking and perfecting, rather than trying huge and costly experiments.
Out to film
For Sundance, we screened off HDCam (a dub from the D5). The movie looked great on a great projector, less so on a lesser projector. We had decided early on that we weren’t going to output to film until we had a distributor, in case there were changes we had to incorporate. (There weren’t, except for logos and company names.)
The last few months have been a process of getting a 35mm release print ready. One of the labs referred to this as a “reverse-DI,” but it’s not really the reverse of anything. It’s the complicated process of trying to calibrate chemical film stocks with digital color look-up tables. I’ve pretty much stayed out of it. The first pass I saw was close enough that I felt certain Nancy and the colorgeeks would nail it.
In most markets, The Nines will project off film, though we may try to use digital projection in certain theaters. The foreign distributor has agreed to use the more expensive Premier (aka Vision Premier) stock for our Venice Film Festival debut — it’s more vibrant, and truer to digital version. Whether or not the U.S. release gets the pricier prints depends on how many quarters are rattling around in the shoe box come late August.
Having seen both the video and film versions, which do I prefer?
I’m honestly much more familiar — and comfortable — with the digital version, so I’m a terrible judge. The film print adds a tiny bit of grain in Part Two, which unfortunately undercuts a little of its video-ness. But on the whole, the film-out looks terrific, and is much less dependent on the quality of the projector. (Why not see both, he asks, trying not to sound greedy.)
I’ll happily answer any other technical questions in the comments. At least, those for which I know (or can plausibly fake) an answer.
- Actually, Part Three workflow is pretty close, but we never dubbed down to DVCProHD. ↩
- We shot Kodak, who did very well by us. Thank you, Kodak. ↩
- Two other cameras did some pinch-hitting: the DVX-100a, and my tiny Xacti camera, which is technically HD. ↩
- Literally, the film has fewer sprockets. Here’s [an article](http://www.digiconform.com/3perf/3perf.html) that explains the difference. You can safely skip the second half, because we never needed to match feet and frames because of our HD post workflow. ↩
- We were an Avid show. I am huge Final Cut Pro guy, but our editor ([Doug Crise](http://www.imdb.com/name/nm0187954/ ↩