On Digital Film Grain and Medium Specificity

I believe that – regardless of the viewer’s breadth of knowledge – one has an inherent sense of what is right or wrong in an image. It comes down to a question of ethics. How a recognisable object – a person’s skin, a tree, moonlit skies, light fixtures – can be rendered obscene when the current filmmaking apparatuses cede control over to the technician instead of the object. All the technical know-how and obsessiveness that can come with the camera (whether it’s a vintage fetishisation of analog or the fastidiousness of keeping up with the latest ARRI ALEXA and Blackmagic models) undermines the necessity and resourcefulness mandated by a limited toolset, a limited toolset more closely resembling the stigmatisms/astigmatisms inherent to vision. It used to be an obsessiveness only affecting filmmakers with access to these tools. But the “prosumer” technocracy now inherent to readily accessible, high-resolution cameras make anyone privy to a once desirable and inaccessible standard, a standard contingent on manipulation (once analog, now digital).

Fixating on a favourite cinematographer also seems the wrong way to approach a film, another tendency in this cinephilia-as-technocracy (or vice-versa). A cinematographer’s artistic contributions – like many a director – can best be approached on a case-by-case basis. In many ideal instances, a cinematographer and director push each other outside of mandates they’ve previously set for themselves, even if it’s only for one or two films. Though ideal, many established filmmakers – of the ilk I mention below – remain ignorant of anything beyond boundaries they’ve set for themselves, resulting in maybe one productive shake-up in their filmography at the hands of a canny collaborator. Viewing a cinematographer’s career standalone, however, can limit their role to that of a visual stylist. There is very little writing about cinematographers that expands this categorisation. Even though it’s not the coverage in itself that’s the problem, it’s what it is premised on, being: the technocracy and fetishization of what is both in and outside of the image.

Of the films below, I watched The Twentieth Century and Uncut Gems at the Lido Cinema and First Cow at Cinema Nova, both in Melbourne. As with most films not finished in a 16:9 aspect ratio, each of the above was shown in open matting, the black bars on either side of their aspect ratios as obvious as the images themselves. This is now the norm in Melbourne, as perhaps only a couple of cinemas accommodate the correct aspect ratios. The projector overcompensates in this regard, the only “fault” of a digital projector that changes how a film is seen.

Watching The Twentieth Century, directed by Matthew Rankin, I was instantly suspicious of the 16mm imagery on display, far too “filmic” despite being shot on film. Looking at an on-set photograph confirmed that it was shot on 16mm, but I got no satisfaction from this knowledge. I still couldn’t decide – while watching the film – whether it was shot digitally or on film. And that still seems like the only valid response to what I saw, regardless of what was actually used during production. I felt much the same watching it as I do when watching a contemporary music video or commercial shot on 16mm. At an earlier point in time, 16mm images transferred to DCP’s or digital files felt easier to identify, but now the digitally scanned 16mm image is ubiquitous to the point of illegibility – either as a fully 16mm or 4K scan of an image. It’s an intermediary likely untouched by human hands, undoing the disingenuous drive to shoot on analog formats, a drive often motivated by bastardised “authentic” or “hand-made” qualities. They may have paid and gone through the effort of shooting on 16mm, but this choice is both too evident as a flourish and too distant from its original format to register as anything other than a stylistic showcase; a choice self-imposed, not necessitated.

I was searching throughout The Twentieth Century for two objects/landscapes that maintain – when shot through a digital sensor – characteristics distinct from celluloid (even when treated with artificial grain and extensive golour grading in post-production):

  1. Practical red light sources: These remain difficult to capture with digital sensors, giving off a pink/paler shade than how the actual light source appears to the human eye. Taillights are an example of this discrepancy in an easily recognisable artificial light source. They often appear in digital photography as a pink-ish glow, showing the components of the light’s redness, but not the colour itself.
  2. Nighttime exteriors: Digital sensors have difficulty capturing deep black hues, especially those of the night sky, unless it’s been treated in the colour grade. Night skies appear as slate gray or dark violet with the clouds often fairly legible. The benefit of this clarity at nighttime is one’s ability to digitally capture the illuminated clouds surrounding the moon, not merely the moon itself – as it tends to appear on celluloid. This is because a digital sensor is more sensitive to natural and artificial light, dependent on a video LOG[1] codex instead of a film negative with a limited ASA. The digital camera also has an ASA/ISO limit, but the ALEXA Mini (the camera used in some capacity on most films mentioned below) has a natural ISO of 800, greater than any existing film stock, which are manufactured nowadays at ASA’s of 500T (T=Tungsten, designed for 500 ASA sensitivity under artificial light) or 250D (D=Daylight, or 250 ASA sensitivity in natural light).

On the subject of self-imposed limitations, the process of shooting on film is now one step closer to a digital camera’s workflow as printing processes, film stocks and laboratories scarcely exist for extensive optical colour timing. The still-enigmatic processes and errors that develop an analog image are the camera’s aforementioned (a)stigmatisms. These (a)stigmatisms are further linked to the optical eyepiece that the filmmaker looks through, as all that separates what one and the camera see are a few glass optics. Inside the digital eyepiece is a screen, one similar to the camera’s monitor displaying its RAW codex of what the camera configures, adjusted with either a customized or ready-made Look Up Table[2]. As such, the digital sensor’s red light source is an approximation of what was previously – between the optical eyepiece and camera negative – a mediation of a recognisable form. There are no inherent (a)stigmatisms to the process of a 2-4K resolution digital workflow. Now that 16 and 35mm images are – with few exceptions –finished within these resolutions, these analog (a)stigmatisms have also been largely smoothed out (though film’s finer-attuned exposure levels still allow for the aforementioned, reconisably human exposure levels and colour temperatures). Each quality in the image is limited to the intentions of their technicians, a megalomania that now dictates how contemporary cinema and – furthermore – the eyes that watch it process these recognisable forms.

On-set photo taken on the set of The Twentieth Century, with Rankin behind the camera

There need to be new, digital limitations. Once-common Standard Definition cameras, constrained to a 720×576 resolution, limit what can be adjusted within a set range according to a lower ISO and higher contrasts. But with 2-4K cameras, the now outmoded techniques of yore (the old lenses and tungsten lights) don’t properly deconstruct the clarity and control inherent to their high resolutions. Anamorphic lenses mounted onto a digital sensor eliminate the time constraints inherent to a 4-perforation film magazine (3-4 minutes on a portable, 400ft mag). But this discrepancy in time – and consequence – changes the image. With these advances made and costs cut, a digitally captured anamorphic image, as a result, looks cheap. This comes back to the intuition I spoke of above. Regardless of the technical know-how or the preference for film or digital, my conviction remains when an image is “wrong”. Digital-sensor anamorphic images are wrong, as is footage treated with artificial film grain in the colour grade, as is celluloid scanned and overly manipulated digitally from the developed negative.

An egregious example of the three tendencies above is in Uncut Gems. Watching it in the cinema, I determined the digital segments from those shot on 35mm, but felt – as with The Twentieth Century – that I’d discovered nothing more of the images themselves. This is the prevailing disappointment of technical investigations and the dubious images that encourage them. There have been too many articles speaking of the technique involved in these dubious images: The Safdies professing their love for cinematographer Darius Khondji, interviews speaking of anamorphic lenses, the time constraints of 4-perf film stock, focus pulling and single-camera setups.

There was a shift from seeing the film on a TV and then at a cinema, obviously, but not in the way I expected. As with The Twentieth Century, open matted black bars were evident, this time at the top and bottom of a 2.39 aspect ratio. What was more evident in the theatre was a discrepancy between footage shot digitally and footage shot on film. In the aforementioned coverage of the film, celluloid was pushed to the forefront as its aesthetic selling point, specifically long lens anamorphic. On Kodak’s blog and in interviews, it was stated that only night exteriors were shot digitally. In the theatre it became evident that the following scenes were also shot digitally:

  • The opening minutes set in Ethiopia (but shot in South Africa)
  • The nighttime exteriors and interiors, which make up roughly 50 minutes of the film’s runtime
  • The scenes towards the end of the film featuring Julia Fox, in both the helicopter and at the Mohegan Sun casino

But again, I felt neither gratification nor disappointment in identifying these scenes as such. Making up nearly half the film, the key identifying feature of the above scenes was the grain predominant in the image. Reading up on it, 35mm elements were filmed on a grey card and superimposed over the digitally captured images, in addition to the high ISO digital noise that appears as its own, distinct grain element in low light conditions. The digitally captured scenes were shot with the same anamorphic lenses as the scenes shot on film, an attempt at medium consistency that, instead, stretched and broke apart the digital sensor, distorting it with an ease non-existent when large-format celluloid is shot through with the same lenses.

The artificial grain and stretched frame felt empty, despite the noise and grain buzzing around the screen, appearing on top of the image instead of within it. The grain didn’t conform to its corresponding colour or contrast, first of all. From an interview with ASC Magazine, Darius Khondji referred to additional grain elements as “grain ghosts”. Taking this into account, one is looking at dead digital images – or at least an image equivalent of an allograft, surgically injected with cadaver bones to compensate for a life non-existent otherwise. The intent of this grain application – as with the anamorphic lenses – is to help it flow with segments actually shot on film. But the overt intent undoes medium specificity. One would shoot digitally in low light scenarios to get a greater exposure level and ability to adjust highlights in the scenes, but the application of grain – both practically and implicitly – undoes this by muddying the actual image that was exposed in-camera. I’m not interested in confronting Khondji on his logic, but there’s suspicion that arises with technicians like him who very easily bow to questions about their technique – most of them encouraged by the likes of both ASC and Filmmaker magazines.

It was in the screening of Uncut Gems that I also saw a trailer for First Cow, a film I’d already seen at home months before. I was instantly averse to the images, adjusting to footage I never remembered watching. As with Uncut Gems, the grain elements were distracting, to the extent that the film’s exposure was also greatly affected by the grain, sapping out highlights and natural light predominant in the images.

Seeing the film again in a theatre (with the black bars, again), I was reassured by a more consistent image, the one I remembered watching at home and whose artificial grain was applied so as to forget about it minutes into the film. Unlike Uncut Gems, it was shot entirely on an ALEXA Mini, creating a more consistent image than Khondji’s mix of analog and digital formats. Kelly Reichardt and cinematographer Christopher Blauvelt also shot Night Moves on the original model of ALEXA with similar amounts of low-light scenarios and grain application.

Colour grading is usually the final step in the filmmaking process, along with sound-mixing, so the trailer’s visual discrepancies may have been from an unfinished grade. Either that or A24 wanted to turn up the levels of exposure, making it more conventionally legible and going against Reichardt and Blauvelt’s intentions.

First Cow (taken from the finished film)

Second First Cow (taken from the trailer)

A more obvious unfinished grade that struck me was in the trailer for Dark Waters. Being the first film Todd Haynes and Edward Lachman shot digitally together, there was a bland uniformity to the images that I don’t associate with either the filmmaker or cinematographer. Watching the finished film, there’s a noticeable difference in the colour grade. The contrast is much more toned down and the image more textured, attributable to the grain applied throughout. Unlike Uncut Gems and First Cow, Lachman utilised a program called LiveGrain, which professes to be a “real-time texture mapping” tool that strategically applies grain to digital footage, advertised as an alternative to shooting on film. Uncut Gems and First Cow applied real grain elements through DaVinci Resolve, whereas Lachman used a more recent tool exclusively licensed to larger-scale productions. LiveGrain’s website shows off a relatively short and exclusive list of productions that have used the software, many of them either Netflix or HBO productions. Because of a lack of transparency on their website, I have no way of identifying a price range for the software nor who could have access to it. Based on the productions that have used it and little else available on the website, it seems more inaccessible than the celluloid it’s trying to emulate – or more likely replace.

Dark Waters (taken from the finished film)

Dark Waters (taken from the trailer)

Reading interviews with Lachman, he expressed resistance towards shooting on an ALEXA Mini, stating that he compensated for the image’s clarity by using Tungsten lights, gels, old lenses and the aforementioned grain software to “deconstruct” the image. I understand his frustrations – likely beyond his control – but shooting a digital sensor through older equipment in no way ameliorates the real (but few) sacrifices actually inherent to high-end digital sensors. Instead of the contrast problems being worked to his advantage, we see it literally covered up in the finished film, an effect highlighting the absence of celluloid.

First Cow and Dark Waters are the least egregious, recent examples of films using digital grain that I know of. But it remains that, through the examples above, I’m no longer interested in arguments of film vs. digital, especially if those advocating for film – like Lachman/Haynes and Reichardt/Blauvelt – are willing to sacrifice the image they’re capturing, so as to sublimate it with the ever-aspirational “filmic” look (a word used to describe everything but images shot on film). Inversely, film itself is also distanced from its material condition by the digital phases required for their editing and exhibition. These methods only seem to be ways of distancing one from their given tools, tools that capture the surrounding light in the little amount of time that permits it, the simplest way I can try to articulate the practice of being a filmmaker.

Through these now-standard practices, filmmaking seems far more distant than it actually is. Even the beneficial immediacy of a digital camera’s images needs to be stretched out in phases and finished in a studio, if we were to go by the films above. The technicians feel it necessary to elongate the process of creating an image either out of habit or by dint of the growing pains still evident in these still-nascent digital apparatuses. The question of a “look” and how it can be manipulated bemuses in the end as – with all the self-imposed and pre-existing hurdles – I struggle to find an image beneath the affects.

[1] LOG (Logarithmic) – The raw, unaffected footage of a digital sensor. Essentially the camera negative, but more neutral and unaffected than its analog alternative.

[2] LUT (Look Up Table) – Used to manipulate a digital sensor’s neutral LOG (seen either through the eyepiece or on its display screen), according to a look decided upon in preparation by the cinematographer and/or director.