I signed out the inspection camera at 09:14. Model: FC-3 FieldCam, second generation, standard wide-angle lens, last serviced 2023-08-02. Unit condition: acceptable; minor scuffing on housing, lens clean.
The assignment was straightforward: document corrosion and structural stress indicators in Sub-Basement 4B of the Wexler Building, a research annex decommissioned after a pipe failure seven years prior. The building has been shuttered since, pending demolition.
My task was to gather photographic evidence for structural analysis. Nothing unusual. Nothing that required more than routine caution.
The corridor leading to 4B felt colder than the upper floors. The air had the stale smell of wet insulation long since dried. A maintenance map from 2008 showed junction boxes along the south wall and a water shutoff valve that had been welded shut prior to closure.
I began photographing at 10:01.
The first anomaly went unnoticed until I reviewed the images later.
The initial problem seemed like user error.
I’d aimed the camera at a small rusted seam where pipe insulation had separated from the mounting bracket. I remembered centering the seam directly in the frame — a textbook shot for the engineering team. But the photo showed it slightly off-center, shifted to the left by maybe four degrees.
A trivial drift. A hand tremor. Poor footing on an uneven floor. Nothing more.
I took another photo immediately afterward, this time bracing the camera with both hands. I said aloud: “Centered. Confirmed.” My spoken notes marked the moment for later sorting.
On review, the seam was again displaced — this time to the right.
I tried to reconstruct the angle from memory. The displacement wasn’t consistent. There was no radial pattern. It wasn’t lens distortion; it wasn’t barrel or pincushion warping. The image simply didn’t match what I’d seen through the viewfinder.
It wasn’t until the fourth photo that I noticed a pattern.
The drift — whether left or right — always shifted toward the far corner of the room.
I resumed inspection, now testing deliberately.
I photographed a safety placard mounted to the wall. Then the fire door. Then a floor drain. Then a spot where mildew had etched faint crescents into the paint.
Every single photo rotated subtly toward the far northwest corner of the room. Always that same corner.
I took a dozen more.
The drift increased.
Two degrees. Four. Six. By the fifteenth photo it was nearly ten degrees.
The camera was not capturing the angle I aimed at. It was adjusting itself, not mechanically but resultingly — as though the recorded image had been nudged before writing to storage.
I rotated the camera ninety degrees, then one-eighty, attempting to “cancel” the drift manually.
The resulting photo was tilted hard to the left, almost useless, and still centered on that far corner.
I wrote the first official note in my pad:
2240-C: Observation 03 Suspicion: field camera orientation not matching optical aim. Deviation appears non-linear; always resolves toward NW corner. Instrument error improbable; evaluate onboard accelerometers later.
At that point, the behavior could still be attributed to hardware fault.
At least, that was what I told myself.
The room itself was unremarkable. Concrete. Four walls. Old conduit. No windows. A single overhead fixture.
The northwest corner was lined with metal shelving coated in dust. Stacks of old inspection forms curled from humidity. An unplugged dehumidifier stood against the wall.
There was nothing notable about that corner.
Nothing that should draw a lens.
But the camera was drawn to it all the same.
To test the optical path, I photographed my own hand at close range. I extended my palm to fill the frame entirely. No background visible.
On playback, the photo showed my hand off-center — and, impossibly, a sliver of the room behind it.
The same corner.
This wasn’t drift. This wasn’t angle. This was an intrusion of something that should not be in the frame.
At 10:41 I switched to thermal mode.
If the camera was being influenced by heat gradients or reflected IR, the thermal profile would reveal it.
I aimed at a conduit clamp near the ceiling. Thermal display: normal.
Captured image: rotated nine degrees toward the corner.
I aimed directly at the floor. Thermal display: normal.
Captured image: rotated toward the corner.
I finally aimed at the corner itself, watching the thermal viewfinder carefully. Cold, uniform. Nothing anomalous.
But on playback, the thermal image registered a faint cold mass — not a shape, just a depression in the temperature field — occupying the middle of the corner at waist height.
The mass wasn’t visible in real time.
Only the recorded image showed it.
To test it, I placed a silvered mirror from my toolkit against the east wall, angled so I could visually inspect the corner indirectly.
I held the camera to the side, photographing the mirror’s reflection instead of the corner directly. The idea was simple: if the lens was being influenced by a fixed spatial point, removing direct line of sight might break the effect.
I aligned the camera with the mirror.
Viewfinder reflection: normal. The corner appeared empty. Sharply defined. Dust undisturbed.
I captured the photo.
On playback, the image showed the corner clearly at last. Centered. Straight. No rotational drift.
But the reflected image contained something else:
A figure — not a person, not a shape so much as a density — standing just outside the reflected field. A distortion where the mirror’s edge should have cut off the view.
Not visible in the room. Not visible in the mirror. But visible in the photo of the mirror.
As if the act of recording had expanded the room by a few inches in a direction the eye could not perceive.
The mirror had shown more than the camera. Or the camera had shown more than the room.
I lowered the device. The overhead fixture flickered once.
I sealed the camera in an evidence bag at 11:03. Two minutes later, the fixture stabilized. The air temperature climbed by 0.2°C.
My final note:
2240-C: Observation 11 Drift resolved when using indirect optical path. Reflection captured spatial detail not visible to observer. Corner may represent non-Euclidean boundary. Suggest quarantine of Sub-Basement 4B pending geometric assessment.
I left the building at 11:17.
As I did, I checked the timestamp on the camera through the bag’s clear window.
It displayed the time as 10:22, though my watch read 11:17 and the device had no external power source.
I did not reopen the evidence bag until the Archive requested this log.
[End of recovered material]