A D.I. Primer

by M. David Mullen, ASC

A “digital intermediate” has become the trendy must-have thing for filmmakers, so much so that they toss around term loosely, trying to sound hip and cutting-edge, making it almost useless as a description of a post-production technique.

“Digital” being a vague catch-all these days, the clue as to the meaning of the term lies in the second word more than the first: “intermediate”. In traditional photochemical post-production, an intermediate is a duplication element (a.k.a. “dupe”) needed to protect the original camera negative from being damaged by over-printing. While shooting, the processed negative is usually transferred to standard-definition video for dailies and offline editing. The final EDL is used by the negative cutter as a guide to conforming the camera rolls to the video edit, cement-splicing the various pieces of negative together. The cut negative is then answer-printed to determine the red, green, and blue printer lights needed to balance the colors and density (brightness) of the image on the print. Using these same printer light numbers, a color-timed positive copy is also made of the negative onto a low-contrast intermediate dupe stock. This is called the interpositive (I.P.); this is then copied onto the same stock to create an internegative (I.N.) from which prints can be made. The I.P. serves as a protection master when archiving the film, plus will be used for the final transfer to video since it is basically a color-corrected copy of your original negative without physical splices. Since the I.N. is on a stronger polyester base, compared to the acetate base of the original negative, you can literally strike hundreds of prints off of it, if not a thousand, before it wears out – and when it does, you just strike another I.N. off of the I.P.

You also need to go through these I.P./I.N. stages if you are converting one camera format (like Super-16 or Super-35) to a standard 35mm projection format. The most common scenario is to take the contact-printed, color-timed I.P. and put it into an optical printer and then re-photograph the image, frame by frame, onto the new film format. You may also be reshaping the image in this process, like when cropping a Super-35 interpositive to 2.40 : 1 and stretching the remaining area to create a 35mm anamorphic (“scope”) internegative.

Just from this description of the traditional photochemical routes, you can see that there is some repetition of labor. You often transfer the film to video twice, once for dailies and again for the final video mastering. Plus you end up color-correcting the image three times: for dailies, for the film print, and for the final video master.

With a digital intermediate, the workflow is a little different – in essence you are creating one digital master that will be used to generate all possible delivery formats: 35mm negative and prints, files for digital projection, plus high-definition and standard-definition video masters. Ignoring dailies, in theory you only color-correct the final image once, and software programs using “look-up tables” (LUT’s) compensate for how the color and contrast will be reproduced when the master is transferred to film or converted to various video formats. The reality is a little more complicated, of course – there is still some need for last-minute adjustments to make these conversions look correct. For example, while the goal is that a film-out from the color-corrected digital master can be printed at one set of printer lights for the entire roll, sometimes a shot or two still need to be adjusted in the printing. Adjustments to the color of the NTSC or PAL downconversions may still be necessary; however, for the most part you are using the original color-correction of the digital master.

Digital intermediate roughly means “film-digital-film”. Some people have expanded the term to include projects shot digitally and transferred to film, but this really isn’t accurate since digital is not the intermediate step, it’s the original stage. Others have expanded the term to include projects shot on film, then converted o digital and projected digitally, which is more acceptable if it means high-end theatrical digital projection, not just showing a standard-def transfer of the movie on an ordinary video projector (which really isn’t any different technically than watching it on a monitor.) Where it gets a bit silly is when people expand the term to include any film-to-video transfer, as in “I did an extreme D.I. for my music video” – when what they really meant is that they color-corrected the image digitally.

What distinguishes a D.I. from an ordinary transfer of something shot on film into a digital video format using a telecine is that the digital master is high-enough in quality and resolution to be recorded back to 35mm negative for making prints. Hence it is meant to be a replacement of the traditional approach of striking intermediate dupes for making release prints, and of having to use an optical printer to convert from one format to another. The truth is, however, that a film-out is currently so expensive that the resulting negative is often considered too precious to make large numbers of prints off of, and an I.P. and I.N. are still struck as protection and printing masters. As the costs of film-outs go down, the original goal of making as many negatives from the digital master as needed will become more common, preserving picture quality by reducing generational loss.

A digital intermediate can be broken down into several common steps:

• Scanning
• Conforming the scans to match the final offline edit
• Adding any standard scene transitions like fades and dissolves, adding graphics like titles, doing any resizing and speed changes, and adding visual effects files
• Color-correcting the edited digital master
• Dirt, dust, and scratch removal
• Film recording
• Conversions to other video formats

Some steps like scanning and film recording are fairly expensive, but standardized at a certain price per foot of film. Other steps like color-correction and dust removal vary wildly in cost estimates because they are based on time spent doing the work, not footage. For example, it is not uncommon to budget ten days for color-correcting a feature, since you are working with a scan of original negative – but at $500 an hour in a DaVinci suite, working eight-hour days, that would be $40,000 total. However, if you could manage to do that work in half the time, you’d be spending only$20,000.

This brings me to another point, which is that a traditional digital intermediate is rarely much below $100,000 for a feature, and often twice that. The film-out alone may be more than $40,000, the color-correction almost the same, and then there are the costs of the scanning, dust removal, and effects to factor in. You have to remember that the stock and processing costs of a feature-length 35mm intermediate by itself is around $10,000 so there’s a limit to how cheap the whole process can be.

But also remember that even if you didn’t do a digital intermediate, should you have to master your film feature to HD as part of your distributor’s video deliverables requirement, you may spend $30,000 or more working a week in a Spirit / DaVinci suite. So some of those high costs of the digital intermediate are covering your HD mastering costs.

Recently there have been ways of reducing the costs of digital intermediates; for example, some people are getting the film transferred to HD and recorded directly to hard drives, or to the lowest-cost DVCPRO-HD format, and do the HD post work, including color-correction, on their own desktop editing systems. And there are cheaper devices than a laser recorder to transfer the HD files to 35mm. I can’t say that the resulting quality is the same as a 2K D.I. done at a professional post house with an experienced colorist working on your project, but it may be a workable alternative if your budget is very tight.

This comes to my other point, which is whether you really need to do a D.I. in the first place. It really depends on whether you need to do digital color-correction and whether you have a lot of visual effects to incorporate into the project, or if you are combining lots of source materials from different formats, such as in a documentary. If your goal is to get a 35mm print into a film festival or theater, it may simpler, and thus cheaper, to just shoot in 35mm in a standard projection format, keep your shooting ratio low, avoid visual effects, and then do a standard negative conform and answer printing. And the truth is that at face value, an optical printer blow-up of a Super-16 feature will be cheaper than a D.I. if your only goal is a 35mm print.

However, if you take the long view that you will need to make an HD master of the movie for home video & broadcast deliverables, and combine that with the likelihood that you won’t get a theatrical release and that you can screen the movie in HD at many top festivals and for private screenings, it may make more sense to concentrate on the digital mastering process and let the 35mm print version be an optional element that you may not need ultimately. Therefore, if you look at the costs of blowing-up a Super-16 feature to 35mm – let’s say that it is near $30,000 – and you combine that with the costs of making a transfer to HD, which may be $40,000, it may make more sense to commit to doing a D.I., because it won’t be much more expensive than that, plus if you end up never needing a 35mm print, you would have saved quite a bit of money compared to making the 35mm blow-up first and then making the HD deliverables.

I also should point out that the D.I. process is not “lossless” –depending on how it is done, your film image may pick-up some digital artifacts and lose some resolution. You may see noise and compression problems, or obvious electronic sharpening, especially if you really alter the image heavily during digital color-correction. The old rule of “garbage in / garbage out” still applies; the best-looking shots will probably be the ones that were the best-looking before the digital color-correction, and the problematic shots will probably remain problematic, although you probably can fix them better than if you only had traditional photochemical techniques at your disposal. But when you see something on the big screen that went through a D.I. and something pops out at you that makes you think “what happened there?”, often it is a shot that needed heavy digital color-correction to fix. So a D.I. is not an excuse for delivering the original photography in a technically inconsistent manner out of laziness.

My last experience doing a D.I. was for the indie feature, “Shadowboxer”. While I originally argued for a D.I. before production as a way of creating a stylized look in an efficient manner, plus as a way of shooting in 3-perf Super-35 and doing the intended blow-up to 2.40 anamorphic digitally, I was told that there was no money in the budget for a D.I. So I opted to shoot the movie using anamorphic lenses in order to avoid optical printer work; for the more stylized scenes, I combined a slant-focus lens with cross-processed reversal stock. It took a lot of light, considering that the reversal stock was only 100 ASA and the slant-focus anamorphic lens was a T/5.6 max – and it would have been simpler just to simulate that effect in post during the D.I. process. Several months later I saw a cut of the movie and it became clear that a D.I. was now unavoidable because it had been edited without regard to doing a traditional photochemical post. About one-third of the total running time had some form of editing manipulation: speed changes, image resizing and repositioning, artificial zooms, repeated shots, etc. It would have been a nightmare to do all of this in an optical printer and the loss of image quality would have been regrettable – a third of the movie would have to be duped. Plus the director had fallen in love with the DVD dailies of the movie, which had been transferred with too much saturation in many scenes. This was something I could only recreate using a D.I. so we searched for a way to make it happen for a reasonable price. We ended up using HDCAM-SR as our digital format, transferring the movie on a Spirit and color-correcting on a DaVinci at Shooters in Philadelphia, a high-end commercial post house.

While using HD as an intermediate step was a little disappointing, considering that I shot the movie in 35mm Panavision anamorphic, using a Spirit as a scanning device had the benefit of smoothing out and “de-graining” our footage –everything from 64 ASA Fuji to 500 ASA Kodak pushed one-stop all looked pretty similar in texture. Also, in those special scenes where I used the 90mm slant-focus anamorphic lens, we had a few shots where I had to use a wider 40mm lens which was not slant-focus, and with Power Windows and Defocusing, I was able to simulate the slant-focus look on those shots to match the surrounding footage. And I was surprised at how good the artificial zoom-ins looked, probably better than if I had done them with the anamorphic zoom lens I was carrying. We were also able to take normal-looking scenes that had been changed during editing into flashbacks and give them a desaturated, bleach-bypassed look. The HD version was then digitally projected at the Toronto Film Festival. It was almost a year later that a film-out was ordered when the movie finally got a theatrical release.

The only downside was the slight softening of the image due to using HD as an intermediate; I knew that a 35mm anamorphic contact print of the original negative would have looked sharper. This problem was only apparent at the premiere screening on one of the largest screens in Los Angeles; for the regular run at the local theaters, I doubt anyone will feel that the image had been slightly softened (and if they did, I could always claim it was an artistic choice…). In fact, the variable quality of scope projection in theaters worldwide will be a bigger factor in how sharp the image will look.

Since that production, I have had mixed success in getting production companies to realize the value in committing to using a D.I. before the filming begins. For example, I have another film now finishing post where we will be doing a D.I. – even though I was told when we began production that there was no money in the budget for one. Again, because we wanted the 2.40 aspect ratio, I shot this movie in 35mm anamorphic just in case it was going to be contact-printed for release, to avoid an optical printer step, but had I been guaranteed a D.I. before shooting began, I could have saved them some money by switching to 3-perf Super-35 and applied those savings towards the digital intermediate.

The excuse you generally hear from indie producers is that they have only budgeted as far as getting a screening print for festivals and distributors, and that when they sell the movie, they can then pay for the home video deliverables. So I thought that once I started dealing with studio productions where the distribution was guaranteed, I could now plan a production around the entire workflow leading to the final deliverables. Wrong.

The real problem is that many non-technical people are in charge of production and their thinking was formed in the pre-digital days where you finished in film and then later did the home video mastering. And many film accounting and budgeting software programs are still built around this workflow. For example, when I suggested to one studio that the savings by shooting in 3-perf could be applied to the costs of the D.I., which in turn would cover most of the home video mastering to HD, I was told that the video deliverables costs were under a different company of the studio and that I couldn’t use my savings in production to get something more in post-production if that included home video deliverables. In other words, the costs had to occur in a certain order, with theatrical proceeding home video, despite the fact that the studio ultimately would be paying for all of it anyway! On another film where the director and I were denied a D.I. by the producing studio, in order to get a digital master from which to make all the theatrical and television trailers, the marketing department spent their own money to strike a 35mm I.P. from our negative before we had even finished timing the answer print – in other words, this half-timed I.P. could only be used once, for scanning purposes in order to make the trailers. Plus I had to take then time during answer-printing to go digitally color-correct these trailers. If we had done a D.I., we could have used the same digital master, with the same color-correction, for the trailers and TV commercials as well as the theatrical prints.

Therefore, making digital intermediates more of a standard element of filmmaking requires re-thinking the traditional cash-flow of a production and recognizing that preparing the video deliverables is as important as making the theatrical prints.

M. David Mullen, ASC received IFP Independent Spirit Award nominations for Best Cinematography for Twin Falls Idaho in 1999 and for Northfork in 2003. His filmography consists of over 30 film titles, including Akeelah and the Bee (2006), The Astronaut Farmer (2006), Solstice (2006), and Shadowboxer (2005).

(Published in StudentFilmmakers Magazine, August 2006)