Signs of synthesis or manipulation

The frame contains no text, lettering, logos, or signage anywhere in the image.

implies

The signal cannot be evaluated; no letterforms are present to assess kerning, glyph repetition, or rendering artefacts.

does not

Anything about the authenticity or origin of the image.

known limit

Establish anything when the frame contains no text, and it is defeated by text that has been inpainted correctly.

No hands or extremities are visible in the frame. The image is a head-and-shoulders portrait with no fingers, wrists, or held objects present.

implies

This signal cannot contribute any observation to the analysis.

does not

Absence of hands does not indicate anything about the image's origin.

known limit

Say anything about a frame with no visible hands, or one where hands are motion-blurred.

Both irises show pale grey-blue colouration with a rendered, uniform radial texture and no visible limbal ring variation. Catchlights in each eye are symmetrical soft-box reflections but lack the micro-irregularity of a physical corneal surface. The lips are closed so no teeth are visible. The right ear is fully obscured by hair; the left ear shows only the lobe area behind the earring, with cartilage structure absent.

implies

Iris texture uniformity and catchlight symmetry are consistent with a synthesised or heavily retouched eye rather than a camera-captured one.

does not

This does not rule out extreme post-processing of a photographic original.

known limit

Work below roughly 100 pixels of face width, or through heavy compression.

Along the right side of the frame, the hair edge against the grey background shows a smooth, continuous gradient with no individual stray strands breaking the silhouette. The transition is uniformly soft across the full length of the hair fall, which is atypical of camera-captured fine hair against a plain studio backdrop.

implies

The boundary behaviour is consistent with a generated or composited hair mass rather than optically resolved individual strands.

does not

This does not rule out an aggressive portrait-mode or AI-assisted background-separation tool applied to a photographic source.

known limit

Distinguish a synthetic boundary from an aggressive portrait-mode cutout on a camera-captured photograph.

The face receives a soft frontal-left key light, yet the skin surface shows almost no specular highlight anywhere on the forehead, nose bridge, or cheekbones. The shadow terminator on the right cheek is unusually hard relative to the apparent large soft-box source implied by the background gradient.

implies

Shadow softness and specular response are inconsistent with a single physical light source of the size the background gradient suggests.

does not

This rules out all photographic lighting setups or confirms the image was generated rather than camera-captured.

known limit

Resolve scenes with several light sources, overcast light, or no cast shadows at all.

Both irises show small, bright catchlights positioned upper-centre. The catchlights are nearly identical in shape and placement in each eye, and neither returns any recognisable environmental detail — no softbox panel, no room structure — that a physical reflective cornea would show.

implies

The specular geometry in the eyes does not correspond to a recoverable physical environment, which is inconsistent with camera-captured corneal reflections.

does not

This cannot confirm the absence of a physical studio setup, as heavy retouching can also erase environmental detail from catchlights.

known limit

Do anything in a frame with no reflective surface.

The face from eyelashes to lips is uniformly sharp. The hair strands at the left and right frame edges, which are at a greater distance from the lens plane, show only marginal blur rather than the progressive falloff expected from the apparent focal length and subject-to-background distance.

implies

The blur gradient does not follow a consistent optical falloff from a single aperture, which is inconsistent with a standard portrait lens rendering.

does not

This cannot separate a synthetic depth map from computational portrait-mode processing on a mobile device.

known limit

Separate a synthetic depth map from a phone's computational portrait mode.

The facial skin across the cheeks, forehead and chin shows a uniformly smooth, matte surface with no visible pore structure, sebaceous variation or fine wrinkle lines at this resolution. The lace garment at the lower right shows floral motifs but the underlying thread weave lacks the irregular yarn twist expected at this scale.

implies

Surface rendering is statistically too uniform for camera-captured skin at this pixel density.

does not

Heavy beauty retouching or skin-smoothing filters cannot produce a similar result in a photographic image.

known limit

Survive heavy beauty filtering, denoising or a low-resolution re-encode, all of which erase the same detail.

The boundary between the hair strands and the grey background along the upper-left shows a smooth, gradient-blended transition with no chromatic fringing and no optical-path sharpening halo. The earring edge against the dark hair is similarly clean with no lateral colour shift.

implies

The edge profile is inconsistent with a camera's optical path, which leaves residual chromatic aberration and micro-halo at high-contrast boundaries.

does not

Post-processing masking or background replacement cannot produce similarly clean edges in a photographic composite.

known limit

Be read on an image that has been re-sharpened after download.

Individual hair strands in the mid-section of the hair fall, roughly centre-left of the frame, show near-identical parallel spacing and curvature across a span of approximately thirty strands, with no stray or crossing hairs breaking the pattern.

implies

Strand regularity at this scale exceeds what physical hair movement and camera capture typically produce.

does not

A freshly flat-ironed, styled hair section cannot show high strand parallelism in a photographic image.

known limit

Flag manufactured objects that genuinely are identical.

The skin across the entire face and neck reads as a uniform near-black with no subsurface scattering, pore variation, or specular highlight gradient that camera-captured skin produces under studio lighting. The light-grey irises sit inside this very dark face with no limbal ring transition consistent with human pigmentation biology.

implies

The skin rendering lacks the layered optical behaviour of physical tissue under directional light, which is inconsistent with camera-captured portraiture.

does not

It does not imply the subject cannot be a person with very deep pigmentation; however, the absence of any skin texture or subsurface variation goes beyond what studio exposure alone explains.

known limit

Rule out an unusual but genuinely photographed event, which is exactly the kind of image people doubt.

The background is a plain, featureless grey studio sweep with no lines, edges, or structures. No vanishing points or architectural elements are present to test perspective continuity.

implies

The background provides no geometric reference to assess perspective agreement or structural continuity.

does not

It does not imply the geometry is correct or incorrect; the signal simply cannot be evaluated here.

known limit

Read wide-angle or corrected lenses reliably, and it is defeated by shallow depth of field.

This case is a single still, so there are no frames to compare.

implies

Nothing. The signal needs at least two frames of the same clip.

does not

That the still is either consistent with capture or synthetic.

known limit

Run on a still image, and it is weakened when we sample only a few frames of a long clip.

High-frequency residual measured over 48 tiles. Median 1.93, quietest 0.85, noisiest 5.00 — a spread of 5.9x.

implies

One region carries markedly more grain than the rest, which is what a paste, a patch or a locally denoised area looks like.

does not

That the region was pasted. Depth of field, sky, and motion blur all produce the same spread.

known limit

Distinguish a spliced region from a region that was simply denoised, and it is close to useless on a heavily compressed file.

No 8-pixel blocking grid is measurable (phase energy varies by only 1.0%).

implies

The frame is either uncompressed, lightly compressed, or has been resized enough to smear the grid away.

does not

How many times the file has been saved. This measurement does not count generations.

known limit

Count how many times a file has been saved, or tell a crop from a paste.

Detail energy across tiles: median 3.74, lowest 0.78 at 75%/0% of the frame.

implies

Detail is distributed across the frame at the level the stated resolution would predict.

does not

Which tool did the resampling, or whether the underlying frame was photographed.

known limit

Tell an upscaled camera photograph from an upscaled synthetic one.

0.00% of pixels clipped to white, 15.95% to black.

implies

Substantial clipping at both ends, typical of a bright physical scene or a hard tone curve applied after the fact.

does not

Anything on its own. This is context for the signals above, not a finding.

known limit

Carry weight on its own; it is context for the other measurements.