Something Did Disappear Over a Flat Surface

1 · What the video gets right

The calculation slide is correct, every line of it. Recomputed independently with R = 20,903,520 ft:

It is better than correct — it is conservative. The slide uses pure geometric curvature; fold in standard surveyor's refraction (k ≈ 0.13–0.17, the “seven-sixths Earth” rule) and the hiding threshold drops further, leaving his lens about 5.2× above it instead of 4.5×. Anyone tempted to answer this video with “you forgot refraction” has it backwards: ordinary refraction strengthens his slide.

The underlying claim is also true, and this site said so before the video existed. Objects really can disappear over a dead-flat surface, with no curvature involved anywhere. On the narrow existence claim he set himself he is right, and he went and demonstrated it rather than arguing about it. The method visibly improves across his three attempts — fixed camera, towed target, thin strings, orientation corrections — which are the instincts of someone actually trying to measure something.

But disappear covers several different things, and they are not interchangeable. A target can fade, losing contrast against its background until there is nothing left to resolve. It can shorten, losing its thin end first as the taper falls below what the lens can render. Or it can be occluded from the bottom up, which is what a hot road does: a heated layer bends the sightline into a false horizon and the car sinks into it. Only the last of those is the ordering the argument needs, because only the last is what a curved surface would also produce — the other two happen to a target sitting in plain view on a flat plane, and would happen to it on a curved one too. Which of the three took the raft is a question of evidence rather than definition, and §2 measures it.

2 · What hid the raft

One observation repeats across all three attempts: the raft is plainly visible from a raised camera and gone from a lens under half an inch. Attempt two isolates it cleanly — the camera descends at a fixed framing, magnification held constant, and the raft goes from plainly visible, to a dot, to gone. That height-dependence already retires two candidates — curvature, which hides zero from every height used, and the raft's angular size, which does not change when a camera rises. What it points to instead is what the vantage itself does to the scene. This is the view from down there:

Which is not the same as the raft being buried in that band, and the difference matters: it is not. Nine-tenths of its height projects clear above the entire pond, silhouetted against the foot of the far bank. Anyone checking the frames will see the raft standing above the water, and they will be right.

Compression does not swallow the raft. It removes any clean background to see the raft against.

Now put the camera in units anyone can feel. Its smallest renderable transition — nine pixels at 0.533 arcminutes each — is 4.8 arcminutes. An eye with 20/20 vision resolves about one. At this framing the video is watching the pond with roughly 20/100 vision, five times coarser than the person holding the phone. A 2.12′-tall, 37′-long, bright-white raft is a comfortable sliver to a human eye; to this camera its entire height fits inside half a blur width, and a pixel can hold only one colour — an average of everything that falls on it. Four inches of raft is outvoted by five hundred feet of scene sharing the same four pixels. A four-pixel object whose entire surroundings fall within one blur width is a marginal detection in any weather. It goes when it goes, and it stays gone.

If that sounds exotic, it is an experience everyone has had. Stand on a beach and there is often a boat out there you cannot make out by eye at all — while the water visibly continues past where it is, so it cannot be “beyond the curve” — and binoculars bring it straight back. The boat was never hidden. It was unresolved, sitting in the crowded band of sea near the horizon where a small angular target is hardest to pick out. This pond is that everyday experience, manufactured deliberately: the vantage squeezed four hundred feet of “sea” into two arcminutes, and the camera brought a fifth of the resolving power of the eye on the beach.

The fade itself says the same thing. Tracking the raft's peak brightness against the background beside it:

The background never moves; the raft dims smoothly where it stands, from full contrast to the noise floor in thirty seconds during which its distance changed three percent. What that does rule out is a mirage line, which replaces what it hides with bright false sky. The background here never brightens.

Which reframes the video's own headline. The raft does not sink; it fades out where it stands. What the frames record is a fade eaten from the outside in — the faint edges drop below threshold first, the bright core lasts longest, and the blob shrinks where it stands until nothing is left. Nothing climbs it; no waterline rises through it. The one directional fact his captions report — the low end going first — belongs to the shape of the lounger, not to any hiding line, for the reasons set out below. To be careful about how far that goes: at four pixels under a nine-pixel blur, a genuine rising cut could not have been resolved either, so these frames cannot separate fade from cut on the raft itself. They do not need to. The two mechanisms that would draw a real cut here — curvature and a mirage line — are independently excluded above, and a tapering target under falling contrast reproduces the ordering with nothing hiding anything. The low end did go first. What that ordering is evidence of is the shape of the raft, not the shape of anything it was floating on.

The three attempts differ in one instructive way

Attempts one and two lose the raft over glass. Mirror-smooth water, unbroken reflections, no wind, no focus trouble — and the raft still dims out as an unresolved dot exactly where the bank, waterline and reflection meet. A white bench a few feet up the same bank stays plainly visible in the very frames where the raft is lost: whatever removes it operates only at the waterline, not on the image. That is the compression account doing its work on a calm day, needing nothing else.

The main run adds weather on top. For its first six minutes the pond is glass. Around t = 380 s ripple trains sweep in; the camera's focus leaves the far field — the treeline goes from its usual 8-pixel edge to 20–22 — and the blurred near water visibly climbs the frame, sweeping across the raft's image band. The raft's contrast collapses inside that window, 415–450 s. The defocus alone accounts for roughly half the collapse; near-field chop, standing into a sightline that begins nine millimetres above the water, is the natural owner of the rest.

Two measurements pin down what that mid-run blur is. It is differential: the treeline's edges roughly double while a reed a few feet from the lens holds its 3–4 pixel edges in every frame — near objects are spared, which uniform fogging, encoder trouble and a shaking mount cannot do. And it is static: across consecutive frames the softened treeline edge holds position to about a tenth of a pixel in the median — under half a pixel in nine columns out of ten, against the 20–22 pixel blur it would have to account for — with no boiling, where air turbulent enough to double an edge width makes it dance. A lens refocused toward the near field — baited, most plausibly, by the newly arrived high-contrast ripple texture — does exactly this. The atmosphere does not.

He anticipates the focus objection with a caption — the trees come back into focus and the raft is still not there — and he is right about that. It establishes that the raft stayed invisible once sharpness returned, which is exactly what four pixels against a nine-pixel floor predicts at full range. It cannot establish why it went, because it went while the camera was soft and the near field was up.

Two more contributors, free of charge

The target itself. The lounger is a wedge — four inches at one end tapering toward nothing — towed on strings, and it yaws on camera; his own method cards say he corrected its orientation mid-run. Thin ends carry less light, so under any uniform loss of contrast a wedge goes thin-end-first, and a yaw toward end-on cuts the visible streak severalfold for free.

But it is worth being exact about what that produces, because it is not the thing the claim needs. A wedge losing its thin end gets shorter, not lower. The loss runs along the object’s length, inward from the tapering end, while the tall end holds. Bottom-up is the opposite ordering: the height falls while the length holds, and a horizontal edge climbs the object with its ends staying where they are.

At this range the two are easy to confuse, and they do look alike — a low object shrinking toward nothing a couple of arcminutes above the waterline reads as sinking, and that is what the eye reports. But the raft’s whole height here is 2.1 arcminutes against a camera whose finest renderable edge is 4.8: less than half of one resolution element, top to bottom. There is no vertical extent available to lose from the bottom. What can change is the length — and length is what the frames show changing.

The null expectation for this object is thin-end-in shortening. That is a real effect and this target gives it for free. It is not bottom-up disappearance, and the two leave different signatures.

The near shore. At the start of the run the raft is filmed through reeds standing inches from the lens — and an obstruction inches away needs almost no height at all: a half-inch crest fifteen feet out hides 4.9 inches at 540 feet. His calm-water check was filmed at the pond's midpoint, where that lever arm is 2× instead of 36× — the one place along the sightline where surface state barely matters.

3 · What it was not

• — Not curvature. Zero inches hidden at 540 feet, from every height he used — his own slide, verified above.
• — Not mirage-strength refraction — and this was the candidate we expected. Bending light the required 80× more sharply than the Earth curves puts the mirage's own horizon 127 feet from the lens: the pond should visibly end there in a band of false sky, the far treeline truncated and mirrored beneath. Checked across all eleven minutes, including every frame of the fade: no band, no inversion, and through every calm stretch the erect reflections run unbroken to the waterline — the signature of a normal profile, not a strongly superadiabatic one. The gust that accompanies the main run's fade cuts the same way, since a breeze mixes such a layer out of existence. This excludes the strength, not the phenomenon — a weak layer shaving a fraction of an inch remains possible, which is what the thermometers in §5 are for.
• — Not the atmosphere in any other costume. Turbulence: in the very frames where the treeline's edge is 20–22 pixels wide, that edge moves between consecutive frames by about a tenth of a pixel in the median and stays under half a pixel in nine columns out of ten — a couple of per cent of the motion that would have to come with turbulent blur. Turbulent air dances; this air does not. Haze: extinction at 540 feet costs about 6% of contrast in ordinary visibility, against an observed loss of 83% — out by an order of magnitude.
• — Not the book's angular-resolution limit. By the book's own 1/60th-of-a-degree criterion the raft should never have been lost: it clears that limit by 2.1× in height and 37× in length. And a resolution limit cannot delete an object in any case — it destroys shape, not presence (every star you can see is far below the eye's resolution limit), it shrinks things toward a dot from all sides at once, and it has no way to take the bottom first. The disappearance lives in the gap between a threshold defined for human eyes and a camera working five times below it — an instrument that cannot test the book's criterion in either direction.

4 · Why a pond cannot speak for the sea

Grant everything the video shows. To bury four inches of raft, something must hide four inches; curvature at 540 feet can hide at most 0.0837 inches, from a lens at zero height.

And local effects do not scale the way curvature does. Curvature's hidden height grows with the square of distance beyond the horizon — 0.08 inches at this pond becomes forty feet at eight miles — while a near obstruction grows only linearly and the vantage effect dies the moment an observer stands up. What survives at sea ranges is the atmosphere itself, which accumulates along the path; that is why long, low sightlines are exactly where mirage effects live, and it is the thread connecting this pond to the two classic low-camera anomalies:

One rule covers all of it: light bends toward denser air. Cool water under warm air bends rays down along the surface, and distant objects stay visible past where geometry buries them — the Bedford Level of 1838, and the Rampion turbines we answer on our own page. Warm surface under cool air bends rays upward, and low objects vanish early — the hot road, and the claim made for this pond. Same layer, opposite signs, decided by which way the temperature runs.

A hull-down ship is a different observation in kind, not degree. Its truncation is resolved: a sharp cut with a sharp superstructure standing above it, at a height that follows d²/2R, unmoved by magnification — zoom shows the cut more clearly and does not bring the hull back. The pond fade is unresolved, smooth, and reversed by optics: where the camera height of his zoomed shots can be established, the zoom recovers the raft. Those are checkable properties of footage, and they are the standard both sides' low-over-water videos should be held to — a P900 pointed at a “missing” skyline is running exactly this test, and what it brings back was never hidden, only unresolved.

5 · The afternoon that settles it

Wallace resolved this exact experimental design in 1870: raise the sightline out of the surface layer, and give the geometry a graduated marker so it can be read rather than argued. The modern version is five short items, four of them possible at the same pond in one afternoon.

• 1 Zoom, from exactly where you are. Same lens height, same raft, focus locked, add focal length: 270 mm equivalent brings the raft to parity with the camera's blur, about 580 mm matches a human eye, and a P900 reaches 2,000. If the raft comes back it was never hidden — only unresolved, a fact about the camera. If it stays gone under five times the magnification, something physical stands in the light path. Either way the largest ambiguity closes in one shot. This is the one test the published footage does not contain: his zooms recover the raft, but every zoom whose height can be established was taken from a raised camera.
• 2 Raise both ends six feet, and film both heights at the same instant. Every mechanism named on this page — compression, refraction, ripples, reeds, and the book's own vanishing-point argument — weakens or dies when the camera rises, so a raft that reappears at six feet localises the cause to the first inches above the water without naming it. A raft that still vanished from six feet would kill them all at once and leave only a plain detection limit — a fact about cameras, not planets. Two phones on one mast make the comparison airtight and hold the background constant.
• 3 Calibrate what hiding looks like. Set a level board in front of the target and deliberately cut off a known amount — at least a foot, since at this framing one pixel is about an inch and anything less is invisible by construction. Nobody has ever filmed what genuine occlusion looks like through this camera at this range; every argument on both sides would benefit from the reference frame.
• 4 Swap the raft for a graduated staff, sized to the camera. A rigid pole ends the yaw and wedge problems, and bands make hidden height a number instead of an impression. Size the bands honestly: the smallest readable band is the blur width times the range — nine inches at 540 feet for this phone — so use an eight-to-twelve-foot staff with one-foot bands, or bring the long lens from item 1. Then walk it out in marked stages and note where hiding begins: a near obstruction hides from the first step; refraction strong enough to matter starts a genuine hiding line around 127 feet; curvature cannot start one until 1,143 feet, beyond the pond entirely. Where the line switches on names the mechanism.
• 5 Instrument the air, cheaply. Shielded temperature probes at 1, 10 and 50 cm, an infrared reading of the water's skin, and a wind gauge — fifty dollars. The footage already excludes a layer strong enough to bury the raft; the probes turn “a weak layer might contribute” into a number, and any breeze above about 2 m/s retires the question on the spot. On ice this item stops being optional. A frozen surface is the classic setting for a strong low-level inversion, so a winter run without a temperature profile cannot distinguish a raft that stayed lost from one that was lifted into view; see the note below and the pre-registration at the end of this section.

What would change our mind

• — A raft that stays gone under 5× magnification from the low position, focus locked. That puts something physical in the path and makes the compression account insufficient on its own.
• — A raft that still vanishes from six feet, filmed simultaneously against a ground-level camera. That kills every mechanism we have proposed, including our leading one, and we would say so on this page.
• — A staff whose hiding line begins near 1,100 feet and grows as distance squared at a rate matching 1/R. We do not expect it inside 4,876 feet, and it would be a genuine surprise.

What would not change our mind is the same run in 4K. To be fair about what 4K is: a cleaner encode of the same capture buys almost nothing, but genuine 4K capture is finer sampling and would roughly halve the camera's floor — the raft goes from four pixels under a nine-pixel edge to about eight under ten. Halving is real; it is also not the tenfold the claim requires, and it does nothing about the geometry, the near field, or the focus. A factor of two where ten is needed, against focal length's five-to-twenty and elevation's everything — the design, not the pixel count, is what the five items above repair.

Written before the fourth attempt

Added 23 August 2026. The analysis above was sent to the author on 20 August, as a link to this page rather than a fixed copy. He has therefore had the five items since before his next run, and he reads this page as it stands rather than as it stood then. That makes the date on this section load-bearing: what follows was written before any fourth attempt existed, and it commits this page to a response rather than forecasting a result.

The distinction matters because the tests above are ours. If a later run adopts one and it comes out as we expect, we will not have predicted his data — we will have proposed an instrument and read the dial. What can be pre-committed honestly is what this page will say in each case.

• — If magnification from the low position brings the raft back, the compression account held — and the run was his while the test was ours. We will say both, and the finding will be about resolution rather than about who was right.
• — If the raft stays gone at five times focal length from a clear sightline with focus locked, our leading explanation was insufficient and this page will say so. The qualifier is doing real work, and it is stated now rather than after the fact: magnification does nothing about a grass stem inches from the lens, so a low run filmed through the near-shore reeds would settle nothing either way.
• — If it is run on ice, the result forks, and we are naming both branches now. Absent a strong inversion the raft should still be lost, because the resolution limit is untouched and dead reed stems stand all winter. But an ice surface actively favours a surface-based inversion, and that bends light down around the surface and can lift distant objects into view — looming, the Arctic case. So a winter run that recovers the raft would not refute anything here, and a winter run that loses it would not confirm much either. What settles it is the fifty dollars of temperature probes in item 5: optional over water, decisive over ice.
• — If none of the five are run, that is not evasion and this page will not imply it is. The tests would remain unrun, which is all that could be concluded from it. People film the experiment they had already planned.

Sources & further reading

• Shape Debate (Levi Miller), Can an object completely disappear over a flat surface? first 3 attempts, YouTube, 19 August 2026 — 15:14, 1080p30; retitled since this page was written, and cited here under the title it now carries: when reviewed it read Can an object disappear over a flat surface? first attempts. Collected by the author at shapedebate.com/flatsurface, a playlist he calls ‘bottom up observations’, which is his own canonical pointer to this material and the one the launch-video description sends readers to; all content as burned-in captions. Source files from the linked Drive folder: three setup clips plus the uncropped 11-minute run (no crop.MOV, 1.34 GB, 1920×1080 30 fps, shot 18 August 2026, 11:07–11:19 EDT).
• Curvature geometry recomputed with R = 20,903,520 ft: horizon √(2Rh), threshold eye height d²/2R. Refraction requirements use the grazing-ray condition; the 127-ft mirage horizon is √(2e/c) at the required ray curvature.
• All frame measurements — fade photometry, edge response, focus timeline, surface-state timeline, frame-to-frame stability, and the attempt-by-attempt vanish sequences — taken from the uncropped run, the camera originals and the published video. Frames reproduced for analysis and comment; each figure states its file and timestamp, and no frame has been sharpened, brightened or otherwise adjusted.
• The turbulence discriminator in §3 is a sub-pixel edge-tracking measurement: ten consecutive frames in each of five windows — a sharp control at t = 300 s and four spanning the fade — with per-column standard deviations of 0.08–0.16 px (median) and 0.13–0.42 px (p90) against a 20–22 px blur. The method is scripts/measure_edge_jitter.py, which extracts its own frames from no crop.MOV and reprints the table. This is a different quantity from camera-aim stability, which is coarser and sampled at 30-second intervals.
• A. R. Wallace, Bedford Level experiment, 1870 — the raised-sightline and graduated-staff corrections adopted in §5. Companion discussion on Bottom-Up Observations.
• A. T. Young, “Inferior mirages: an improved model,” Applied Optics 54(4), B170–B176 (2015) — the appearance of real inferior mirages: the vanishing line and false-water band that §3 checks for.
• H. R. Blackwell, “Contrast thresholds of the human eye,” JOSA 36, 624 (1946) — detection as a contrast-and-flux threshold, distinct from resolution; and H. Koschmieder via W. E. K. Middleton, Vision Through the Atmosphere (1952) — the contrast-transmittance figure behind the haze estimate.