Lighting Lies: Flash, Top Light, and Grain

Flash makes grain look coarse

A flash fired from a handset a metre away returns a hard specular band and doubles the apparent depth of every pore on a leather panel. Readers comparing that frame with a diffused frame of the same item report a texture difference far larger than anything the material itself shows. The recorded effect is consistent: direct on-camera flash makes grain read roughly one grade coarser than it is. The reader is not being careless; the photograph is supplying contrast that the surface does not have. One grade is the unit used throughout this dispatch, and it means the visible step between a fine grain and a medium one at the same viewing distance.

The distortion has a signature that can be counted rather than felt. Flash frames carry a small highlight with a hard edge somewhere on the panel, a shadow falling directly behind raised texture, and a falloff that darkens the frame corners by more than the centre. Under this light a leather sample with pores spaced about 1.2 millimetres apart will read alongside a 1.6 millimetre sample without the two being distinguishable. Count the highlight, the shadow direction and the corner falloff, and the light source has named itself before anyone discusses the material.

Movement between frames is the cheapest available test. Two photographs of one panel taken seconds apart under flash show the highlight in nearly the same place, because the light source travels with the camera. Two frames taken beside a window show the highlight moving as the seller shifts position. A highlight that never moves across a six-frame set marks an on-camera source. Where the highlight moves in two frames and holds in a third, the set was shot under mixed sources and should be split into two separate readings.

Grain is not the only property affected. Flash flattens saturation on dark panels and exaggerates it on light ones, which is why the same bag can look greyer in one frame and warmer in the next. Edge paint and plated hardware take a harder highlight and read sharper than they are. None of that makes the frame dishonest; it means the reading has to name the light before it names the finish. A reader who sorts by light first will discard some frames entirely and never argue about them.

Why the error happens

The physics is ordinary. A small source close to the subject produces near-parallel light with a hard terminator, and the terminator sits exactly where surface relief does. Diffused light arrives across a wide angle, fills the troughs between pores, and reduces the contrast that an eye reads as depth. A photograph is a contrast record, so the flash frame is not lying about the material; it is reporting the lighting. Nothing about a seller's intent can be inferred from this, and the desk files it as a method difference rather than as a concealment.

Telephone cameras add a second stage. Most compress the image and then run a sharpening pass that raises local contrast along edges. On a coarse surface the edge network is dense, so sharpening amplifies texture selectively. On a smooth surface there is little to amplify. Two panels cut from one hide can therefore separate in a single frame purely because one sat closer to the lens. The same mechanism exaggerates stitching and print edges, so a jersey photographed at fifteen centimetres shows a sponsor print reading harder than it does in the hand.

Distance compounds it. A handset held ten centimetres from a panel covers a smaller area with the same flash output and produces a brighter, harder frame with more visible pore structure. The same handset at forty centimetres produces a dimmer, softer result. Sellers photograph what is convenient, and convenience usually means close and lit by the device, which is the pairing that distorts texture furthest. Asking for a frame at thirty centimetres or more, with the light arriving from the side, removes most of the error without asking the seller to buy anything.

Video calls make this worse. Most conferencing tools apply a smoothing filter that reduces low-contrast texture while leaving strong edges intact, so a screen-shared inspection can show a panel looking smoother than the still frames taken minutes earlier in the same room. Where a seller offers a live walkthrough, a still photograph from that session remains the safer basis for a decision, because it can be measured afterwards and the call cannot. Motion also blurs single frames, and a blurred frame is the worst evidence available for texture.

Diffused daylight practice

Ask for a frame taken near a window, out of direct sun, with the item at least thirty centimetres from the lens. The window becomes the light source and the falloff across the frame stays gentle. Two frames should be supplied: one square to the surface and one at roughly thirty degrees, because a raking angle reveals warp and lifting that a flat frame hides. The request names three numbers, which is deliberate; requests that name a number get answered, and requests for better photographs do not.

A reference object belongs in the frame. A ruler, a coin, a bank card or a printed grid serves two purposes, giving the reader a scale for texture and spacing while proving the frame was staged rather than lifted from a catalogue. Where a seller supplies a reference object without being asked, that is worth noting in the record, though it is not evidence on its own. The object only becomes useful when it sits in the same plane as the surface being judged; held toward the lens it measures nothing.

Repeat shots matter more than resolution. Three frames of one panel taken within a minute under the same window let a reader separate real variation from capture noise. When a mark appears in one frame and not in the other two, the mark is usually a reflection or a sensor artefact. When it appears in all three at the same position and length, it is on the item. Resolution adds detail to a bad frame without settling whether the detail belongs to the panel or to the capture.

Write the light into the note. A record reading diffused window light, two angles, reference card present can be compared against a later frame from the same seller, and a change of method then shows up as a change rather than as a defect. Records that omit the light force the next reader to start the whole reading again from nothing. The desk keeps the method line short enough to type in one gesture, because a field that takes effort to fill gets skipped.

Open the finds catalogue

Consequence comparison

The cost of reading flash texture as real texture is a refusal that should never have been issued. In the recorded sample, buyers who rejected a leather item after seeing only flash frames described the grain as coarse, and the same buyers re-reading diffused frames of the same item described it as normal. The item did not change; only the light did. A refusal on texture alone is therefore the weakest refusal available, and the easiest one for a seller to answer with a second set of photographs.

The opposite error costs more. A smooth-reading frame produced under heavy diffusion can hide weave separation, lifted edge paint or a shallow crease that a raking highlight would have shown plainly. Readers who prefer soft light for every judgement lose the one angle that reveals lifting, so the desk recommends both methods in sequence: soft light for texture, angled light for structure. Reading structure first also stops a reader from spending an exchange on grain while a bond is quietly failing.

Time is the third cost. A lighting dispute usually consumes one exchange with the seller and sometimes a second set of photographs, which pushes dispatch back by a day or more. Naming the three conditions in the first message removes most of that round trip, because the seller either holds the frames already or takes them once. Where the item has already been dispatched, the same three conditions become the basis of an arrival comparison, and the frames taken before dispatch are what a claim will be argued against.

Records kept per item should carry the lighting method beside the verdict. Two months later the only question that matters is whether two photographs are comparable, and the method line answers it where memory cannot. The desk treats a verdict filed without a lighting note as provisional and marks it that way in the log. Provisional verdicts are re-read whenever a second frame set arrives, and they are never used to justify a refusal on their own.