ACQUISITION FORMAT AND "LOOK": ISOLATING VARIABLES
Steve Yedlin
Camerimage: Bydgoszcz, Poland
November 15, 2015
-------------
Intro
Hi there. My name is Steve Yedlin. I too am a cinematographer and I'm humbled and a bit overwhelmed to be here amongst so many distinguished and passionate imaging professionals. Today I'm going to share a sliver of my own personal philosophy on how aesthetic look emerges from technical processes.
By "look," I of course mean that elusive concept that we purposely and usefully keep vague so we can use it differently to suit different situations. But the general idea is an overall perceptual feeling from an aggregate of many image attributes that might make one image feel different from another even if they were both photographing the same thing under the same light.
Clearly, this elusive high level aggregate quality emerges from low level physical processes, but do we mere mortal cinematographers have any control over the image at this lower level or are we forever stuck manipulating large turn-key systems that must be taken in their entirety? Is it possible to isolate variables so that we even know which attributes matter and then manipulate those attributes separately?
And so, I'd now like to present a short demo I've created on the subject...
--------------------
The Main Talk
Thanks for watching. I've been so nervous and excited to share this with you all.
I'm guessing some of you would like to discuss the demo, and indeed the key aim of this presentation is to provoke a dialog, but before we start a discussion, I want to take a few minutes to elaborate on some things that there weren't time for in the short demo and that are on a slightly more philosophical vector.
First, I'd like to reinforce what I said in the demo that I'm not claiming that that the two images you just saw are indistinguishable or identical in every aspect, because they wouldn't need to be for my thesis to carry weight -- concentrating on that is a distraction.
The thesis of the demo is merely that a camera itself is only a data capture device and doesn't necessarily enforce an indelible stamp on the substance of the overall look of the final image. If a camera system acquires enough data, any visually desirable attribute can be identified, modeled and reproduced using mathematical transformations in post. Preparing the collected data for display in any desired fashion is possible. The fact that this usually isn't done is not proof that it can't be done. I don't believe that there are attributes of the perceptual "look" that mystically persist and are immune to all transformations.
I'm not talking about tiny clues that can be used to forensically identify the acquisition format. I'm talking about the overall look.
In the demo, I used my own personal model of some of film's attributes which is thorough enough to satisfy me... At least for now. Someone else might care to model the same attributes more precisely or in a different way or to model other attributes of film, or of course to create a mathematical model for any number of looks that are not at all related to photochemical acquisition. Or to make an inverse transformation.
My hope here is to pave the way for us to jointly recalibrate our current lexicon, which tends to be based on entrenched doctrinaire beliefs and community-reinforced certainty but not on empiricism.
Like a religion, we were born completely immersed in a presupposition that's not questioned despite the lack of evidence to substantiate it. We've inherited the belief that a camera format has an unalterable signature look -- that that thing we call "look" comes from the capture format and is immutable. A belief that one physical machine at the beginning of a long chain of processes is solely and irrevocably responsible for the brunt of what we see at the end of the chain. We believe in it so thoroughly that we even believe this look shines through heavy color correction.
We're all familiar with phrases such as "celluloid just has that soul" and "you just can't get that look from ones and zeros" and "digital has that clinical look." These phrases make a point of expressing a sweeping value judgment while failing to identify any specific attribute that's responsible for the "feel" or the "soul" or the "look."
They express our insistence that a camera format enforces an indelible character to the image's overall look and that whatever attribute by which it does so can never be altered, understood or even identified.
These sentiments also fail to consider whether a historically documented difference in look when using different capture formats is inherent or if it's a result of repeating the same processes, which are not necessarily the only possible options.
But I believe that if there is any difference (that's not mere illusion) between any two looks, then it is certainly not beyond our means to identify those attributes and then to model them. I don't think it's beyond science or human ingenuity to identify what attributes make two images look different from one another and to make mathematical models of the physical processes.
So I'm urging for us to break free of the current dreary cycle: the feedback loop where the predominant belief expresses a certainty that attributes are elusive, mystical and can never be identified or modeled, so we refuse to try. We cite the results of not having done it as proof that it can't be done. Clearly a self fulfilling prophecy if there ever was one.
And this goes well beyond the topic of film versus digital or the "film look," which is really just a small example of a larger issue. This applies across the board to how we think about the perceptual results of the physical processes of imaging. We speak of all sorts of mechanisms as if they're monolithic holy grails rather than tools -- specific models of digital cameras, specific numbers that appear before a letter "K", or certain bit depths of image files, and so on.
So, my hope is that, together, we can strive for a liberating and exciting paradigm shift in our lexicon and thought process for understanding how aesthetic looks emerges from technical processes in photographic imaging.
I'm optimistic that such a perspective shift is not only possible but can lead to an enlightened transcendence and new freedom that allows us to cast off the weight of a perceived duality that's bogging us down: just as physicists who were on the cutting edge in their day but seem naive in hindsight debated bitterly and confusedly that light couldn't be particles because it was waves. They eventually realized that it was neither the wave side nor the particle side that was wrong: it was the dogmatic absolutism itself that was wrong. And the wave/particle duality had to be accepted as true and non-paradoxical.
So too I hope that we imaging professionals can overcome the current doctrinaire skirmishes between one-sided camps. Whether it's film-versus-digital or red-versus-alexa or 4k-versus-2k or any other: let's remember that these are not monolithic deities at whose altar we must worship, but maleable tools at our disposal that can be mastered in ways we haven't yet fully realized.
Thanks for your time. That concludes the diatribe portion of the presentation. I also want to show you a few more details that are specific to the demo, but I'd like to open it up to discussion first.