Just Because You Can, Doesn't Mean You Should.

/

Just Because You Can, Doesn't Mean You Should.

April 29, 2015

or "Contemporary Digital Imaging Problems"

More specifically, my own problems with contemporary digital imaging; problems that are totally subjective, largely aesthetic, and a byproduct of evolving technology. The characteristics I've singled out are commonly found in photo and video all over social media, the web, TV, movies, and in print. It seems even a few famous photographers and cinematographers have a bit of a heavy hand with their digital images. For professionals and amateurs alike, the aesthetic difference in photography before and after most people in the world began carrying a camera around in their pocket is a massive one. I fear I may be pointing out the obvious but perhaps not. Bear with me on this unintentionally longwinded tirade.

The image above is of some elderly men playing mahjong on a rare sunny afternoon in Beijing. It qualifies as a photograph as it was captured with a camera and lens but through poor use of image processing software, it's been voided of nearly all its photographic qualities. I used a filter here actually, a rather expensive product from Visual Supply Co. (VSCO) that was supposed to make it look like it was shot on Kodak UltraMax 400 film. However not only does it look nothing like UltraMax, it now bears little resemblance to the scene it reproduces and has become something more akin to digital illustration than photography. To me it looks like a cartoon. These days I wouldn't release an image like this but getting to this point has been a long process of trial and error. Along the way I've unintentionally butchered a good number of otherwise decent photographs.

In the past five years there have been major advancements in all aspects of digital image processing, HDR (High Dynamic Range) in particular. Various forms of this technology have found their way into low cost consumer devices and the highest end professional equipment. This has contributed to the collective movement away from the photographic qualities of camera-based image making. In the current paradigm, an image's characteristics are no longer determined solely by photographic parameters but as much by how scene data is captured, encoded, and then interpreted by software. We've been on this trajectory for quite some time but In my opinion, it's plausible the evolving role of software in imaging could eventually push out traditional photography completely. 

The traditional photographic parameters are:

The choice of optic and its specific qualities.

The use of in-front-of-the-lens or behind-the-lens filtration.

The choice of film stock and its specific qualities.

How this medium is exposed to the light in the scene.

The technique of photochemical development.

And then varying amounts of red, green, and blue light through the developed capture medium onto a final delivery medium.

Along with this limited set of variables comes fewer possible visual treatments for photographic images. There are also limits to how much picture information can be usefully captured onto emulsion and how much it can be altered after the fact using only photochemical tools. These limitations actually provide the primary aesthetic distinction between what photography was and what photography has become.

In modern digital image capture, data describing the scene can far exceed the limits of human vision. Almost every last little scrap of detail in the brightest and darkest parts of the frame is recorded and this comprehensive information can be interpreted by processing software an infinite number of ways. Some of these possible treatments will be deemed aesthetically pleasing whereas others will not.

For now, because the memory of what film actually looks like is still fairly fresh, digital images that exhibit film-like qualities seem to resonate more positively than those without.

The qualities of film images are:

The grain characteristics of the emulsion. Even low ISO stocks designed for sharpness are sharp without being hard edged or "graphic". Separate layers of emulsion for red, green, and blue light, each with a different photochemical reaction never perfectly overlap. The pleasing softness and natural quality of these subtle imperfections has yet to be precisely emulated in digital.

While it varies from stock to stock, hues and tones tend to be reproduced in a way closer to how we experience them with our own eyes. Though it's a subjective and visceral reaction, a more human level of contrast and saturation gives images a quality of truth and authenticity.

I'm not too focused on the motion characteristics of 24 frame per second film, depth of field in the 35mm format, etc. These problems have already been solved. I'm far more concerned with somehow finding a ring of truth in something inherently lacking it; a closer analog to our own visual experience.

Digital images are:

Often too pristine, too perfect, overly sharp and lacking texture, sometimes having a strange plastic-like smoothness.

Specific to the manufacturer, each digital camera has its own identifiable characteristics that can be hard to conceal. Though it's a great system, it has a very particular look that whenever I see it I think, "there's another TV show shot on Alexa with Master Primes." Because of the popularity of this format in particular, to my eye a lot of contemporary content looks the same.

High dynamic range can be too much of a good thing. While having access to a scene's full tonal range is helpful, particularly for composting and VFX, heavy handed use of it is the primary destroyer of otherwise good images. When excessively bright and dark values are digitally remapped closer to the middle of the tonal scale, it looks wrong because it's antithetical to our own visual experience. Our eyes just don't see in HDR. When it's bright, our iris closes to let in less light causing our brains to perceive less detail in darkness. Predictably, the opposite occurs in the absence of light. Most research agrees we perceive more or less 10 stops of tonal range at once. Anything beyond this requires the eyes to adjust. Most digital cameras now capture 12, 15, 18 stops, or more by blending multiple exposures. HDR TV, the latest electronics industry gimic, supposedly displays 17-20 stops. In demonstrations I saw in 2014, it looked hyperreal and weird and is in my opinion, another step in the wrong direction. Digital 3D fizzled, 4K TV's have undersold, so perhaps this latest thing will entice consumers to buy yet another TV. 

Whenever I think of technology unintentionally butchering photography and filmmaking, The Hobbit: An Unexpected Journey in HFR 3D (High Frame Rate) comes to mind. Technology has opened a lot of new avenues for creativity but it's also created a daunting sea of possibilities where it's far easier to make bad decisions. The near universal opinion on HFR 3D is it was not a good use of the technology. A noble technical effort but the effect was bizarre so routinely rejected by audiences. 

I currently see very little difference in digital imaging for still photography and motion picture filmmaking. While the toolset varies from software to software, they all essentially do the same thing and the skill of the operator is now as important as those in the field doing the capture. In my career, I've attended online sessions for movies and TV shows where I've seen colorists basically re-light entire scenes. When digital images are perfectly exposed, high range clipping and noise floor can't prevent post production artists from pushing or pulling the image in any direction they choose. The fact that authorship can now so easily be taken away from its creators is problematic in and of itself. The results can be believable but upon scrutiny, have no photographic truth simply because the photography had a different intention.

Examples of all that I've outlined can be found anywhere people share photos and videos. I would never call out someone else's work and as I've been guilty of it all, I'll use my own work to illustrate.

The image below is a raw capture from a Sony A7R, a remarkable little camera that records very high resolution, high dynamic range files. The lens used was a Leica Summicron 35mm, an extremely sharp optic that I've found occasionally looks a little hyperreal on a camera with this much resolution. This image is problematic because it's extremely high contrast but by capturing at a low ISO and exposing for the highlights, for better or worse, every last little scrap of detail has been preserved.

My processing software examples are Adobe Lightroom but most of what I'm suggesting readily applies to any image processing application. The toolsets are different but the way pixels are affected is largely the same across the board.

Through a lot of trial and error in Lightroom, I've come to the conclusion that if you want to ruin otherwise decent photos, start with "White and Black Clipping" and "Presence". Six out of seven of these tools adjust the image in a way that would be photochemically impossible or at least highly unconventional. It's very tempting to use these global adjustments to make the sky a little less bright or people less silhouetted but it doesn't take much before the image loses its photographic qualities and takes on the heavily affected look of HDR. I now stay out of here completely.

VSCO and Alien Skin sell Lightroom presets that attempt to emulate a wide variety of film stocks. VSCO offers six different filter packs with hundreds of film stocks and processing options. It's an overwhelming amount of choices and after spending way too much time experimenting with them, I've found not one looks anything like it's supposed to. VSCO has provided some aesthetically pleasing examples on their site but I personally haven't found value in any of these filters and have spent a lot of time and money to arrive at this conclusion.

A small selection from VSCO 5, "Archetype Film Stocks", 189 filters in total.

The image below is VSCO's Kodak Royal Gold 400 emulation. Having shot hundreds of rolls of this stuff in high school and college, this filter looks nothing like Royal Gold. It just looks weird and wrong. If you want the look of Royal Gold, shoot Royal Gold and do it for the very same reason the world's preeminent filmmakers have agreed to purchase enough stock to keep Kodak in production. 

This is AGFA Vista 400 but it has the same problems as the Royal Gold filter, mainly an artificially compressed range of contrast.

This one is supposed to look like a color reversal film, Fuji Velvia 50. Can this image even be called a photograph anymore?

With access to every last little bit of picture information, there's nothing stopping the photographer from bringing the highlights down, opening up the blacks and shadows, adding a little pop of color and voila, Looney Tunes. Look familiar? iPhone HDR + Instagram filters has yielded many results like this.

Without using VSCO or digging into unused dynamic range, an image that's more photographically plausible can be achieved. The contrast is extreme enough that without digitally cheating, we're not going to hold on to everything so I've let the highlights go. If this were a professional shoot, a large silk or net would be used to control this contrast and bring the highlights down to a more acceptable level. Doing this digitally can't emulate the interactive qualities of light so will never look as good as the real thing.

The only thing saving this image is the high resolution of the A7R as the problematic parts of the frame can be cropped out. Perhaps this version is even more compositionally and thematically interesting. 

One last example; below is a raw capture of a "street doctor" in New Delhi, India. Sony A7R with Leica Summicron 35mm. This photograph is inherently a little flat and lifeless.

The first time I processed, the image below is how I released it. I now realize what went wrong. After applying contrast, I felt the shadows were too deep and the highlights too bright. After opening things up and clamping things down, it still looked a little lifeless so I added a touch of color, sharpness, and warmth. In trying to alter the inherent qualities of this photograph, I unintentionally ruined it.

Applying what I've learned the hard way, in the image below I've taken out the digital sharpness, let things go dark and bright where they're supposed to, desaturated a touch, left the color temperature balanced, and added a little texture. Is it now a good image? That's not up to me to decide but this treatment offends me far less than the previous example.

In both my professional and personal work, every time I've felt like I finally wrapped my head around this I've soon found myself back at square one. When there are infinite possibilities, accessing an image's aesthetic potential can be an arduous process. For my own tastes at least, I've identified the qualities I aspire to and the pitfalls that prevent it. Moving forward, I'm striving to keep in mind the simplicity of photochemical photography and its qualities; the texture and softness of it, the faithful reproduction of tones and hues, and most importantly, accepting imperfections and resisting the urge to endlessly tweak. Therein lies the root of this problem; in the effort to enhance, it's far too easy to lose an image's authenticity.