Up until now, modded reality has been a mostly external affair, but recently, I got to wondering what it would be like to peer out on the world from inside my own body. Surely such a bizarre experience would be worth investigating. So I decided to do what any normal person would: mount a camera in my mouth and stream the video to a VR headset.
The resulting experience is a uniquely oral outlook to be sure, although one that is perhaps not the most practical, even by modded reality standards. Just constructing a device that goes into the body was challenging, and the result can best be described as an interesting failure.
Still, just watching myself eat green jello from inside my own mouth was worth the price of admission. It was gross and wonderful all at the same time. And while experiencing the world by mouth may sound regressive at first blush, I found that the experience connected me to my body in a whole new way.
Camera and Electronics
My goal was to mount a camera inside of my mouth looking out on the world, preferably far enough back so that my front teeth and entire mouth opening would be visible, and then stream the video from the camera to a VR headset. Sounds simple enough perhaps, but I ran into a few difficulties that prevented me from fully realizing this vision.
For the camera, I selected a cheep USB endoscope from DBPower. The camera body itself is a tube about five centimeters long and one centimeter in diameter. This makes it the smallest, cheep endoscope that I could find, but, as I soon found, it was still far too large for this particular application.
The camera works out of the box with Video for Linux and on Windows using DirectShow. Here’s the list of supported capture formats:
$ sudo v4l2-ctl --list-formats-ext ioctl: VIDIOC_ENUM_FMT Index : 0 Type : Video Capture Pixel Format: 'YUYV' Name : YUV 4:2:2 (YUYV) Size: Discrete 640x480 Interval: Discrete 0.033s (30.000 fps) Size: Discrete 160x120 Interval: Discrete 0.033s (30.000 fps) Size: Discrete 176x144 Interval: Discrete 0.033s (30.000 fps) Size: Discrete 320x240 Interval: Discrete 0.033s (30.000 fps) Size: Discrete 352x288 Interval: Discrete 0.033s (30.000 fps) Size: Discrete 800x600 Interval: Discrete 0.067s (15.000 fps) Size: Discrete 1280x720 Interval: Discrete 0.200s (5.000 fps) Size: Discrete 1280x960 Interval: Discrete 0.200s (5.000 fps) Size: Discrete 1280x1024 Interval: Discrete 0.200s (5.000 fps) Size: Discrete 1600x1200 Interval: Discrete 0.200s (5.000 fps) Size: Discrete 640x480 Interval: Discrete 0.033s (30.000 fps)
My previous experiments all targeted either 60fps or even 120fps, so a cap of 30fps was a bit disappointing. Thirty frames per second is also the absolute lowest limit of what works with VR goggles, which meant the best resolution I could hope for was 640x480.
As for the camera and optics, the field of view is very narrow, which is not uncommon for endoscopes, although it made this experiment significantly more difficult. Unless the camera is literally in the back of your throat, there is simply no way to capture the entire interior of the mouth, and the best I was ever able to achieve was having most of the upper and lower incisors in the frame. And because the camera has to be mounted so far back in the mouth, you end up viewing the world through a fairly narrow window.
Speaking of the world, endoscopes are designed to image objects at fairly short distances, and the DBPower model I’m using has a depth of field of between around one and four inches. This works out well for imaging features inside the mouth, but not so well for viewing the world outside. Objects at any distance are blurry, but generally recognizable enough, at least when the camera is operating in ideal conditions.
Despite the change of cameras, I’m using the same streaming pipeline as in my previous experiments, with the endoscope connected to a Raspberry Pi running mjpeg-streamer. An iPhone connected to the Pi accesses the mjpeg video stream over LAN, and the phone is used inside of a Google Cardboard headset.
The Raspberry Pi is powered from a battery, so the entire getup is portable with around two hours of battery life. Since I wasn’t planning to actually take this device out and about, for portability, I just stuck the Pi in my pocket.
After quickly tossing together the basics, when I fired up the camera for the first time, things quickly came to a stuttering halt. At a resolution of 640x480, forget about 30fps, I was seeing something closer to 5fps with around a second of latency. Not good.
top revealed the cause:
Whereas I had previously used mjpeg streamer to stream two 800x600 streams at 60fps – with under 50% peak cpu usage—now, this measly little 640x480 stream at 30fps was pegging the CPU, leading to dropped frames and increased latency. What gives?
Well, looking at the endoscope’s supported formats, notice that it only supports YUYV data. Mjpeg-streamer, as its name suggests, creates an mjpeg video stream, so it must convert the YUYV data to mjpeg for streaming. And that’s the main difference from previous experiments. The USB fisheye cameras I used previously could capture mjpegs directly, which made the entire pipeline many times more efficient.
Unable to get mjpeg streamer performing well at 640x480, I reevaluated a few other streaming tools, including ffmpeg, vlc, and gstreamer, but was unable to achieve anything close to what mjpeg-streamer provided in terms of low latency and ease of use. So I decided to stick with mjpeg-streamer, but drop the resolution to a paltry 352x288. This is unfortunate, but it allowed consistent streaming at a blistering thirty frames per second.
And honestly, while a resolution of 352x288 at 30fps looks really bad on paper, especially compared to what a Vive or Oculus offer, please do remember where this camera will be going. In that context, resolution and frame rate are hardly primary concerns.
Sticking a camera in your mouth is more difficult than it may sound. Sure it’s simple to just hold a camera between your lips, but that lacks a truly oral outlook. Similarly, there’s no real difficulty in playing dentist and just shoving a camera into your mouth, but such an inward looking perspective proves impractical for navigating about the world.
No, for this experiment, I was after a more particular view: a view from inside my mouth looking out on the world, with front teeth, cheeks, and tongue all in the frame. Such a view requires that the camera be positioned in the back, center section of the mouth, which is not the most convenient of places to put a camera.
My first mount attempt used armature wire:
The larger sections on either side are held in the cheeks, with the connecting wire passing behind the back most molars. The camera is strapped in the middle and faces forward. Shaping the wire to fit comfortably without busting up all my teeth required some trial and error.
The main problem with this mount is that the endoscope I am using has a comparably narrow field of view. Therefore, to get any oral features in the frame, the camera has to be mounted fairly far back on the armature, which itself is already at the back of the mouth. Although the camera body is only five centimeters long, the device ends up extending into the throat, inducing a terribly unpleasant gagging sensation and occasional retching reflexes.
In the end, I found this Jaws-esque mount too uncomfortable and cumbersome to use. Maybe I could get more used to wearing it and even train myself to suppress my gag reflex, but it’s somewhat absurd that just wearing this little device brought me far closer to exorcising than any of my previous, highly disorienting experiences in modded reality ever did.
I also experimented with mounting the camera to a mouthguard. Starting with a soft rubber mouthguard molded to fit my mouth, I used a highly precise bonding technique to construct a state-of-the-art, oral camera platform using the latest biomaterials (read: heating paperclips with a lighter and poking them into the three dollar plastic mouthguard).
The mouthguard fits on the upper teeth and is more comfortable to wear than the metal armature previously detailed, yet, since the camera still extends into the throat, it also suffers the same gagging problem. Also, after trimming the mouthguard down in the front so that it would not show up in the view, I found that it no longer fit snugly and can come loose easily.
The mouthguard camera still didn’t offer the mouthscape I was hoping for, so I returned to the metal mouth armature, this time turning the camera sideways and mounting a small mirror at a 45 degree angle in front of the lens. This allowed me to instead mount the camera sideways in my mouth, using half of the metal armature first detailed.
Because the camera no longer extends into the throat, this mostly eliminated the gagging issue. It also introduced further complications, not least of which is that it is difficult to actually center the camera in your mouth using this mount.
My mouth is about five centimeters wide—as measured by the outer distance between my back molars—and, if you recall, the camera body is also five centimeters long. Those numbers just don’t add up. With a good chipmunk cheeking, it should be simple to find the extra few centimeters required to position the camera in the center of the mouth, but the camera is position too far back to actually extend into the cheeks, and instead runs into the much more solid and unyielding jaw muscles. I tried several different designs to get around this.
None were terribly comfortable or really fixed the problem. Moving the camera slightly forward allowed me to better center it but eliminated my ability to actually bite down.
Furthermore, take an already shitty endoscope camera and point it an even shittier mirror (I swear that thing is little more than cardboard covered in silver Sharpie) and the result is not pretty. Even outside the mouth, the image quality is poor at best, and, once inside, the mirror fogs up at the merest exhale. The mirror doesn’t even cover the entire view.
Still, this mirror mount did provide the best view, so I had to make do.
Although this device is extremely basic, building it drove home the difficulties of creating devices that integrate with the human body. Size and choice of materials become incredibly important considerations, and my preferred tape and glue prototyping techniques only sort-of worked.
Even with these difficulties, the mouth is by far the most accessible and easy to work with internal space. I didn’t have to worry much about sanitization or biocompatibility, and neither did this experiment require the slicing, blood, and pain of a true grinderian implant.
Looking out on the World From Inside My Body
When I started this experiment, I gave little thought to the practicality, or even the desirability, of viewing the world through one’s mouth. It sounded like it could be fun, and, to my knowledge, no one had really done it before so I figured, why not?
Well, spoiler alert, although the experience was certainly unique, I can’t say I entirely recommend it. The ad-hoc nature of the device is partially to blame, but there’s something more to it as well.
First off, as detailed, all the camera mouth mounts I tried were, at best, uncomfortable to wear. Perhaps the best was the camera plus mirror setup, with the metal wire held in one cheek. This mount offered the best view of my mouth’s interior and was easy to adjust. It also made me gag the least (high praise, that). The main downsides were that I could not clench my teeth together while using it, and that the image quality is terrible at best.
After getting past my initial reluctance to stick the decidedly inedible camera into my mouth, I positioned the camera for what seemed like the best possible view, before slipping on the VR headset and entering the strange world of inner space.
With the camera held on the left side of my mouth, I found that my front, right incisors and canine teeth were all in the frame, along with the left side of my mouth opening.
At least in my humble opinion, the inside of the human body is generally rather gross looking, and the mouth is certainly no exception. Sure, lips are well and good enough, and a wealth of fine Mardian pearls is hardly unattractive, but those are mere visages. And, moving a touch more inwards, while there’s no denying the pleasures of exploring someone else’s mouth through passionate kissing, that operation is more tactile than visual, with the actual oral cavities in question remaining happily out of sight, and all the biological happenings therein remaining happily out of mind.
But this camera… it’s more of a dentist view than anything else, with teeth, cheeks, gums, and other bits all rendered in unflinching closeup. The whole scene is a good deal too fleshy for my tastes. My lips and the walls of my cheeks specifically were constantly pulsing and twitching all about in a terribly concerning manner. It all was very raw and wet and red, and I doubt that anyone could stand being in such a place for long.
The teeth were also a cause for concern. My teeth certainly aren’t perfect, but I thought I maintained them well enough. Looking upon the ivory train wreck before me however, I really started to doubt my flossing competency. Surely, the only fix for a dental disaster of this proportion would be to rip everything out and start fresh (yet another reason why it’s good that I went into software instead of medicine).
Peering out at the world from between my lips, shapes and objects were blurry and indistinct but generally still recognizable. At first it was difficult to remember that I had to keep my mouth agape to see the world, and it also took time to learn that I had to move my entire head just to look about.
A bigger problem was breathing. Because of how the Cardboard was strapped to my face, breathing entirely through my nose was difficult. But just the slightest exhale through my mouth instantly fogged up my vision and left me blind. Inhaling again cleared away some of the fog, but soon saliva started collecting on the mirror and camera lens. This took me from blurry but usable varisoft style vision, to something more akin to peering through frosted glass blocks.
But worst of all was the limited field of view, which I first really noticed when I stuck one of my arms out. Whereas with the fisheye lenses I used previously, most of my arm would have been in view, even with my arm fully extended, only my hand was ever in my vision. This made it difficult to see what was going on anything closer than two feet from my body, a skill which is actually really important when trying to grab something from a table or just open a door. Because the view is so narrow, I also had to turn my head about a lot just to get a sense of the entire scene about me.
This more telephoto style vision also flattens out depth. I never got a good sense of how far away anything was, which is one of the reasons why I look much more tentative in the videos while walking about or interacting with the world.
The mouthguard based camera offered a far less immersive picture of the interior of my mouth, but a much sharper view of the outside world. Normally, only my upper teeth and upper lip were in the shot, although the mouthguard also offers a fine tonguescape view as well.
Using either device, I could close my mouth, but I found that I was unable to effectively swallow. This led to saliva pooling around my tongue, and, because I had to constantly keep my mouth open to see anything, this in turn made tilting my head forward, even the slightest, an altogether drizzly operation. I couldn’t even safely look down at my feet to see what was in front of me.
Closing and opening my mouth provided no end of interesting sunrises and sunsets, yet I couldn’t help but feel a tad dirty watching the days go by. Of course, anatomically speaking, the view doesn’t match any normal human organ or orifice—at least those I am familiar with—but, like a piece of H.R. Giger’s work, it does seem to capture some abstraction of those ideas. Combine that with the teeth and gums in the foreground—not to mention that obscene tongue wagging all about—and you have yourself some real horror movie type stuff.
And, like all truly great horror movies, beyond the initial gross-out factor, I think there’s something in this experiment that is disturbing on a deeper level. It touches on some of the same feelings that I experienced when using my heartbeat to modulate my vision, namely, confronting one’s biological entirety.
Sure we all know the experience of owning and operating a mouth, but rarely do we pause to consider what is actually going on down there. Take eating. For me at least, the actual sensation of eating is mostly decoupled from the biological act. From my perspective, food goes into this magical hole in my head and it makes my brain feel good. And while I’m at least somewhat aware of all the chewing and swallowing and other housekeeping involved, the overall experience is all from a more indirect and detached perspective.
The view this device offers however, strips away any of the mouth’s magic and abstraction, leaving it little more than a gaping maw, the tail end of a digestive tract leading from the anus. From my new perspective, I could very well have been looking out at the world from inside a chimp’s mouth, or from inside a dog’s mouth, or from inside any of the enumerable mouths found in the animal kingdom. It’s one thing to know that you are merely flesh, but quite another to be faced with the ugly and undeniable truth of the matter head on.
The two potential applications of a mouth camera are, of course, gormandizing and oral sex. Being a decidedly boring individual, I declined to investigate the latter, although I cannot envision any variety of that activity working well using this device, if from nothing more than a mechanical perspective, nor do I imagine the resulting experience being at all desirable for anyone involved, although tastes vary of course.
As for eating, my culinary experiences were limited to an infantile diet of liquids and soft foods. The camera mount hinders chewing, and swallowing required all sorts of tongue movements and full body gyrations. And while normally I’m good at keeping my various internal juices securely inside my body, while eating wearing this device, I kept leaking saliva and partially processed bits of food everywhere. This was highly embarrassing. It also makes eating rather unpleasant to watch for the observer, but a thoroughly entertaining experience for the eater, provided at least that you give up all pretense of dignity.
I began my culinary journey wearing the mirror camera mount. This mount offers a good view of the action, but prevented me from closing my teeth together. Therefore, my choice of foods was very limited. Pastes, such as pudding or apple sauce, tended to get stuck on the camera and make me go blind, and are also just not that interesting to watch being processed. Better results were had with jello, which is solid enough to somewhat resemble real food, but soft enough so as to not be too dangerous.
The actual experience of watching myself eat from inside my own mouth was certainly unique, but, again, not altogether desirable.
Perhaps the most humorous view came before the meal even began. Seated at the table and looking down through my mouth at the bowl of green jello cubes before me, I couldn’t help but laugh. Because the camera was mounted on my head, the view wasn’t that unfamiliar, but now there were these crazy teeth and gums and the foreground (fun fact: by the year 2019, most photos of food will be taken from a mouthward view, with the real taste makers having moved on to esophageal and gastric food photography).
Scooping up some jello with my spoon, I carefully navigated it towards the center of my vision. With the jello before my lips, light glinted through it like so many emeralds, before I boldly slurped into the unknown.
I found that I could choose between politely eating with mouth closed and being effectively blind, or eating with mouth fully agape, like the most uncouth of beasts, but able to see the world. I chose the latter, because it’s not like someone was filming the whole thing or anything.
Eating may also be too strong a word for the procedure. Since there was no actual chewing involved, most of the process consisted of using my tongue to squish the jello cubes into more manageable chunks and guiding them around the camera to the back of my throat. While this was going on, the jello hopped all about, moving in and out of view like so many bits of flubber.
Then came swallowing. First off, I had to get over my instinctive reluctance to even try swallowing while this device was in my mouth (quit your sniggering). The device’s location near the throat also disrupted normal food swallowing procedures.
What I found worked was tilting my head backwards to collect the food in the back of my throat. With gravity now on my side, using a precise series of gyrations, I then eased the jello down into my esophagus, which took over digestive duties from there. Needless to say, the entire operation is not the most attractive and would hardly go over well on a first date. And, in the auditory department, my performance can best be described as mix of choking to death, retching, and moist sloshing (coincidently, the avant-garde opera I am working on consists exclusively of those three sounds).
This whole procedure worked well enough for larger jello chunks, but I was never able to entirely clear out my mouth. Food kept collecting around the camera or in pockets in my gums, and then dribbling out in a most embarrassing manner. A bib is highly recommended.
And then, somewhere around my third spoonful, suddenly I went blind. Turning on the camera’s LEDs revealed the cause, a mass of jello had lodged itself in my new eye!
Visually, jello vision is actually kind of neat, like stained glass almost. I tried unsuccessfully to clear up my vision using my tongue, before groping about for a glass of water. Swishing this around cleared off the camera, but then the water started dribbling everywhere. Even with my mouth closed, the water still managed to leak out around the cable, and swallowing only partially cleared things up.
I also tried eating using the mouthguard camera, which was much more practical and actually allowed me to somewhat chew my food. The image quality was also much better. The downside was that the view of my mouth was not nearly as immersive; more like looking out of your head at mouth level, rather than having your eyes truly inside your mouth.
Overall, while I expected eating to be rather gross, and it most certainly was, the entire operation was so exquisitely ridiculous as to be amazing. Maybe watching myself eat more traditional foods would inspire me to give up eating all together—either that or I’d end up choking to death; in the end, the result would be the same either way—but you just can’t get all existential while noshing on green jello and spilling it all over yourself like a proper infant, even if you are watching the whole gross procedure from inside your own gross carcass.
While there’s undoubtedly interesting possibilities in taking modded reality into inner space—some of which the biohacking community has explored—the technology for seeing the world through one’s mouth isn’t quite there yet, at least for the home hacker on a thirty five dollar budget. Other types of sensor may be much more practical and easier to get started with, but devices that actually go in to the body—as opposed to external sensors that monitor internal function—bring numerous additional challenges and concerns.
Still, while viewing the world through my mouth didn’t work all that well, it was a fun and thought provoking exercise. Perhaps one day soon, when technology catches up, experiencing reality from inside one’s own body will be common :) And what a future it will be! You’ll be able to digest yourself and shit yourself out, or even give birth to yourself! Sadly though, as my personal loadout did not include a womb, I’m limited to the former option. Being “born again” could be fun though. One day, my dear Nicodemus. One day…