Taking Inventory of Yourself

Photogrammetry, Control, and Identity

tywen kelly
9 min readJun 9, 2021
A photogrammetry portrait made from 30 self-portraits taken over 30 days.

The Process

At the start of February 2021 I began to take an image of myself everyday. I sat down on the edge of my bed, straightened my posture, and looked forward at the same spot in the corner of the room. Each day I moved the tripod a few inches to the right along a semicircle marked by tape I had placed on the carpet. I would set the shutter on a timer and position myself in the dictated spot. I would pre-visualize how the soft Rembrandt lighting would hit my face as I waited for the flash to pop. I did this for 30 consecutive days.

The purpose of taking these photos in a consistent manner was to ease the post-production process. I was going to feed these images into photogrammetry software and create a composite sculpture made up of the 30 self portraits. At around the same time, I had been playing with photogrammetry as an artistic tool to understand how images-as-material can generate a gestalt “mass image” and how this mass can have ontological presence and weight in our lives. The series of photogrammetrical sculptures I was producing — everyday things like flowers, houses, motorcycles — channeled Nonhuman Photography, or an aesthetic of the distanced, miasmic, outward-facing gaze so persistent in computer-mediated image making. My thesis was that nonhuman photography has a weight in our lives, though it is often invisible: things like thumbnails, proxy files, and CCTV footage are rarely foregrounded. I felt like cosplaying as a Seeing Computer, as Trevor Paglen describes it, and scan the world through the eyes of a computer.

But in attempting to channel the nonhuman vibe, I discovered it was impossible to emulate that gaze in totality. I was taking photogrammetry image data sets by hand, using my camera as a paintbrush to coat the object with images, intuitively letting my camera go to parts I found exciting rather than areas that I had had not covered. Most of my scans were left with holes, imperfect replications, rife with glitches: proof of subjectivity. The kind of photogrammetry I was conducting was very non-nonhuman photography.

The path of images maps the movement of the camera in my hands as I held down the shutter. It winds and circles back. Joyously inefficient.

Accepting this barrier is what inspired me to flip the photogrammetry camera on myself. Instead of emulating the outward-facing gaze of the tool, I would turn it to face inwards to the thing making the image datasets. Perhaps then I could really understand my non-nonhumanness.

While running the thought through my mind I began to think of the new possibilities I had access to after turning the camera around. For one, I could take images of myself more thoroughly than I could with any other subject. Most of the objects I was scanning I only had access to for a few minutes. I would scan the tree, bush, or fallen log, then leave. It was a photojournalistic capture, raw punctum with no studium, that isolated the captured object into a clean vial. But I had access to myself for more than a few minutes. Hours, days, and months! Could I capture my non-nonhumanness?

A “light stage” can capture photorealistic models. (Source)

Photogrammetry is picky with what you feed it. Classical photogrammetry datasets are captured consecutively and as quickly as possible to reduce any unexpected behavior from the software. Any variation in the subject’s movement, scene color, image’s spatial compression (via focal length), and exposure between images causes glitches. A common subject of photogrammetry experiments are rocks, because they are inanimate things with low reflectivity and no translucent components (reflectance and translucence often cause aberration artifacts in the final 3D model). Many scientist-artists in pursuit of photorealism have created elaborate camera contraptions, such as large geodesic domes made up of dozens of camera pointing inward, which scans an object in mere milliseconds. An example of this is USC ICT’s Light Stage used by Hollywood to procure cinema-quality assets.

Each image in a boomerang time lapse from the 30 day period.

Unlike a rock, I am an unideal subject. I cannot sit still. There was a great deal of variation during the process of shooting the self-portraits. My hair moved, my facial expressions shifted, my mood swayed. Invariance was the only constant. After compiling the images into a photogrammetry mesh I noticed a few things: The mesh edges were frayed, melted, and glitched where the software had incorrectly deduced where my ears and neck lay. My face was recognizable but altogether alien. My skin bumpy. I could see that the photogrammetry software tried hard for realism, but ultimately it could not reconcile the affordances of day-to-day living. But this was not unexpected or bad. In fact I resonated with it. What I immediately saw in this futile attempt to amalgamate my 30 self-images was a bit of myself.

Control of Identity

Identity formation is, as I understand it, a cycle of breakage and reconstruction. What I was observing from my portrait was that photogrammetry works in similar ways. Perhaps there is some lesson to learn from that.

What is identity? Identity is not a rock. It makes and re-makes itself. Every new day adds an image to your greater mass of self images. The next day the new image is instituted into the mesh, requiring a complete rewrite and re-composition of your identity. This rewrite becomes the foundation for the following day, when more imagery is added. And so on. The accretion of daily identities is an additive process. Each image is not a sedimentary layer in your identity, clearly demarcated from other images. It is mixed in like a viscous fluid encoded in between the mass of images around it.

Memory works in the same way as identity formation. Memory is an active reconstruction of past events. For nearly a century, researches have theorized that when one “retrieves” or remembers a memory, they are not plucking it off shelf as if it were a static object. Instead, memory is more like thousand bits of a shattered mirror that you must piece together each and every time you recall it. To recall is actually to reconstruct; and reconstruction is a projection of biases, frameworks, and schemas. Similarly, our identities undergo a continuous shattering whenever we interact with others, reflect on our histories, or confront our biographies — they are delicate things. To make sense of our identities requires habitual maintenance and reconstruction.

The UV texture map for the photogrammetry model, fragmented into islands, shattered! Without any formation of it, it looks like a mess.

Digital avatars, online MMOs, BBSs, timelines, and feeds have become the source of the greatest identity shattering day-to-day. As such they have also become the set and setting for identity formation. The virtual characters we inhabit on-screen are increasingly entangled with our real personalities. Researchers have observed that we start to behave in real life depending how we are made visible online — this is called the Proteus Effect.

The Proteus Effect was a study that found that people who played a video game (Second Life) and were given an attractive avatar were not only more likely to act more confidently and extroverted in-game, but also in real life for a period of time afterwards. The same research group, the VHIL at Stanford, confirmed this effect in later iterative experiments which placed participants in a VR headsets and put them in scenarios where they were forced to act heroically and found that after the participants took the headset off they acted more altruistically in real life.

The notion that our identity is influenced by external images of ourselves is intuitive. What may be less intuitive is creating a conceptual framework to understand how to deal with it. After all, it is overwhelming to understand that our identities are shattered every second by the fast-paced metaverse. Our identity seems out of our control. Here, it is useful to point to Gilles Deleuze’s concept of “dividual”, a portmanteau of divided individual, to describe this anxiety.

We no longer find ourselves dealing with the mass/individual pair. Individuals have become “dividuals,” and masses, samples, data, markets, or “banks.”

Deleuze’s critique cites a societal shift away from a method of top-down control which sought to mass individuals into a vague group as a means of control. Before, a factory owner may seek to remove individuality from members of a group they want to control — like laborers — by lumping them into a “mass” of people, diluting their perception of individual power. Today, individuals are hyper-siloed and given what me might call a “digital twin”, or a statistical profile composed of discrete data points that describe the person, as a means to track and subsequently control their deep psychology.

The crux of Deleuze’s anxiety is that our dividual selves are in the hands of someone else. He does not point to an inherent evil of a the concept of a self divided into two; what is evil is that the statistical profile is generated invisibly in the background and out of an individual’s reach. In an essay titled Inventories, Not Identities Kei Kreutler speculates on a future of “web3”, or a cryptographically-based web age, which is an internet that not only generate copies but can also generate unique “non-fungibles”, and where a person’s identity becomes entwined with their assets. It’s a future in which one’s crypto wallet becomes your identifier — the assets you own (or have owned) cryptographically form your face.

Is this an… inventory?

At first glance this seems like a late capitalistic dystopia, but upon closer inspection it placates some of Deleuze’s anxieties of a society of control. In web3, a user could own their dividual identity, or at least their public facing one, and control their profile. Even “shadow profiles”, or metadata that is generated from your behavior (or non-behavior) to create profiles of imaginary members, could become public is a radical two-way transparency. If everything is tracked — even the trackers — then what happens is sudden and complete ownership of oneself. Making sense of the tracker means having an image at hand of the process that is shattering your identity.

Taking an inventory of oneself is thus a means of identity formation in a metaverse that shatters your identity, byte by byte. A Quantified Self approach so to speak in conjunction with an omnipresent sousveillance web3 blockchain is a technosolution that can exist alongside other methods. Regardless of the means, however, the problem is the same, which is that to control our identities requires that we can make visible the controlling mechanisms that form them.

Photogrammetry and Identity

What do the metaverse, the Proteus effect, and dividuals have to do with photogrammetry? They serve as the background and context in which photogrammetry operates. Photogrammetry can visualize the system of control in the air and serves as a powerful metaphor for identify formation today.

Wearing my generated portrait as a mask.

Taking a self portrait of myself every day was like making an inventory of my identity. Each image was a performative decision for that day to remain similar to the day before, or to break character. Each image became a reenactment of how we self-generate a new image of ourselves every day. I had to ask myself: Do I be like yesterday? When I took those photographs and ran it through photogrammetry software, it projected onto the list of my portraits how an alien (as in a non-me) intelligence would understand it. As Susan Sontag noted about photography, the camera “projects” the photographers biases on the subject, more so than it captures the bias of the subject themselves. The final photogrammetry model was a piece of evidence, an act of taking inventory, which served as a map of the messy act of forming identity.

“Owning” a photogrammetry portrait means more than using photogrammetry. Owning it is an exercise in media literacy and understanding the invisible processes. Owning it is reveling in the glitches. Owning it is seeing the tension in creating fleshy forms from flat images; in seeing the tension between creating continuous identities from discrete chunks. Owning it means engaging with the nonhuman systems of control to make it visible and non-nonhuman-readable. When we own it, then we can have more control, or at the very least, learn to be at ease with the lack of control we have.